Fix a latent bug in the Python wrapper that was triggered in pybind11 v2.6.1

These changes ensure that the GIL is acquired before calling functions in C++ that
may consume a Python object with reference count 1 as an argument, in which case
the object's destructor is called at the end of the function, after any GIL acquired inside
the function would have been released.

Fixes #1271.

* Fix pybind11 issues.

* Retract suspicious fix.

* Address comments.

Replace CMAKE_SOURCE_DIR with PROJECT_SOURCE_DIR to pick up the correct paths when used as a sub-module.

Fixes #1266.

Major changes in Python library and documentation:

* Add global state (immutable after initialization) for the Python module that
  manages the mapping between probe metadata types and the method by which the
  corresponding [sample data can be collected and presented as Python objects.
  This is necessary for decoupling the implementation of the simulation wrapper
  from that of the various cable cell probe types.

* Wrap each of the C++ cable cell probe types with a Python function that
  returns a `probe_info` object.

* Add code for converting and accumulating C++ probe metadata and sample
  data, registered in the module global state against the metadata type info.

* Make the `arbor.simulator` object responsible for recording all spike and
  sample data, with corresponding new methods.

* Use NumPy arrays and structured datatypes for returning spike and
  sample data.

* Place Python schedule proxies under an abstract interface so that consumers
  of the proxies can be made generic over schedule types.

* Add unit tests for the `arbor.simulator` class and its new functionality,
  including distributed spike collection.

* Rework all Python API documentation concerning probes, sampling, and spike
  recording; add information on newly exposed cable cell probe addresses.

* Add new python example `single_cell_recipe.py` which is a generic recipe
  version of `single_cell_model.py`.

* Adjust other code in the wrapper and examples to accommodate these changes.

Minor changes in Python library:

* Add equality tests for `arbor.location` and `arbor.cable` Python classes.

* Use the `py::` namespace shorthand more often in the Python wrapping sources.

* Add an implicit conversion for a pair of numbers to `arbor.cell_member`, so
  that e.g. probe ids can be specified simply as `(gid, index)`.

* Add an implicit conversion from a tuple to `arbor.mpoint` so that a
  morphological point can be specified simply as `(x, y, z, radius)`.

* Modify the interface to `arbor.regular_schedule` so that the overloads
  correspond more closely to the C++ API. In particular,
  `arbor.regular_schedule(dt)` now describes a schedule with times 0, dt, 2*dt,
  etc.

Minor changes in C++ library:

* Change test in FVM lowered cell implementation and exception message for
  `bad_source_description`: assert number of detectors is exactly the number
  of sources promised by the recipe.

* Add comment describing `probe_metadata` object in `sampling.hpp` which
  asserts that the type and value of probe-specific metadata completely
  determines the correct interpretation of sampled data provided to a callback.

* Return `arb::morphology` from `load_swc_xxx` methods for consistency with the NeuroML interface. 
* Fix unit tests.
* Fix documentation

Speed up plotting in seaborn.relplot by disabling calculation of confidence intervals

* Remove blank line from `ball_and_stick.swc` 
* Mention that blank lines in SWC files indicate end of data.
* Check for trailing data after parsing in the python interface.
* Leave stream in good state if `parse_swc` encounters an empty line indicating end of data.

* Rename `as_segment_tree` to `load_swc_arbor`.
* Remove `swc_mode`: `sort_and_validate_swc` now performs the `relaxed` checks by default, `load_swc_arbor` performs the `strict` checks. 
* Change `load_swc_xxx` to only accept `swc_data` as an argument
* Remove default constructor for `swc_data` and enforce validation of SWC records via `sort_and_validate_swc` at construction.
* Add fixes for docs and unit tests.

* Add interface for querying names.

* Add derived mechanisms to list, clean-up.

* Add unit tests.

* Add python bindings and docs.

Add a catalogue of mechanisms used by the BBP models.

* Add some log messages in python examples
* Add python examples to CI

* Python examples: Updated to use region expressions (quotes)
* Python examples: Uniform plotting
* Python examples: Removed tutorial directory (example roughly matches python/example/single_cell_model.py)
* Python examples: Rewrote flat_cell_builder() to segment_tree()

* Synced pages between Concepts, Python API and C++ API wherever possible
    * Recipe pages conform between the three section (concepts, python, c++)
    * Cell, Cable Cell and Cell * pages are rearranged and provided with some copy explaining the relationship between them.
* Moved Python API out of Concepts
* Renamed Concepts "How does Arbor work?"
* Added Python Module Index plus mock import of Arbor for RTD build (unfortunately won't show there)
* Broke out Interconnectivity (synapses) page.
* Reworked Single Cell Model page into a quick start, with lots of cross referencing.
* Tweaked logo.
* Added Spack to install options.
* Updated blurb.
* Documentation now follows EU capitalization rules.
* Assorted typofixes

* Minor output formatting fix for `swc_record`.
* Modify the Python `load_swc_allen` implementation to cope with SWC
record ids not necessarily being contiguous, and with SWC record parent
ids corresponding to record ids, not 0-based indices.

* Update `dryrun` to generate some spikes again, add a voltage printer.
* Fix `ring.py` recipe.
* Fix labels in `single_cell_builder.py` and `single_cell_swc.py`.

* Simplify scanning of SWC records.
* `parse_swc` now retains metadata read from SWC comments.
* `parse_swc` is given two modes, relaxed and strict. Strict mode enforces contiguous record numbering and rejects SWC that looks like it has a single-point soma.
* Subtype `swc_error` according to error mode.
* Remove `point` class and corresponding unit tests.
* Add a RAII stream formatting flag saver, used in SWC record output operator.
* Replace included 1-point soma SWC example with a regular SWC from NeuroMorpho.

* Substitute `std::optional<T>` for `arb::util::optional<T>` for non-reference types T.
* Remove reference-deducing `util::value_by_key`; replace usages with new function `util::ptr_by_key`.
* Add some missing header includes that were required but included only transitively.
* Remove `operator<<` overload for optional in python/strprintf.hpp; replace with utility wrapper class that catches `std::optional<T>`.
* Wrap some `std::optional` values with `to_string` in ostream output in python lib.

Fixes #1154.

* Move label and s-expr parsing code and unit tests from Python to arbor C++ library.
* Update `locset` and `region` constructors that take string arguments to parse strings as s-expressions or "quoted" labels.
* Modify the input stream modifier used to parse asc files to use a lookup table for substitutions, renamed it transmogrifier.                                                             
* Replace `hopefully` type implemented in python headers with an `arb::util::expected`.
* Add `ARBDEV_COLOR` CMake option that forces gcc and clang to always output color output.
* Allow arbitrary strings in labels in region and locset expressions.
* Add `parse_region_expression` and `parse_locset_expression` functions alongside the existing `parse_label_expression` function for use when a region or locset is expected. These calls will promote a quoted string `"label"` to `(region "label")` or 
 `(locset "label")` respectively.
* Add user-defined string literals for labels so that the C++ interface can use `"soma"_lab` instead of awkward escaping `"\"soma\""`.
* Simplify Python wrapper code.

* Replace usages of `util::any` with `std::any`.
* Modify `util::any_cast` so that it forwards `std::any` arguments to `std::any_cast`; move into own header.
* Prevent `std::unique_any` from implicitly encapsulating `std::any` objects.
* Replace `test_any.cpp` unit tests with `test_any_cast.cpp` unit tests.
* Replace some usages of `std::is_same<X, Y>::value` with `std::is_same_v<X, Y>` as part of general C++17ification.
* Remove `util/variant.hpp`, `test_variant.cpp`.
* Replace occurrences of `util::variant` with `std::variant`.
* Replace `util::variant::get<N>` with `std::get<N>`.
* Replace `util::get_if<N>` with `std::get_if<N>`, passing a pointer to variant instead of a ref or value.
* Add some standard #includes where they were required in the source, but only included transitively.
* Some minor comment tidying and fix-ups.
* Update C++ API docs.

Fixes #1144.

* Make the members of `cable_cell_ion_data`  optional.
* Add separate methods for painting `init_int_concentration` , `init_ext_concentration` and `init_reversal_potential` on the cell.

* Implement a workalike for the proposed `std::expected` class: see http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2019/p0323r9.html .
* Replace use of `either` with `expected` in `mprovider`, `mechanism_catalogue`, `util::partition_range`, and `pyarb::hopefully`.
* Replace use of `either` with `variant` in `util::sentinel_iterator`.
* Add `in_place_t` constructor for `util::optional`.
* Fix move assignment bug in `util::variant`.
* Remove `util/either.hpp` and associated tests.

Fixes #1135. 

* Update CMake CXX version.
* Make CUDA 10 a minimum requirement and remove CUDA 9 support code.
* Set up g++ Travis tests to use g++ 8.
* Explicitly set C++14 version for nvcc.
* Properly split cuda compilation, including in unit tests.
* Remove unnecessary modcc `SOLVE` warning.
* Update pybind11 module to tag v2.5.0
* Replace `util::size` and `util::data` with `std::` equivalents.

Switch `arb::util::either` to `arb::util::variant` in the Python wrapper.

Fixes #1133 

This PR adds a bunch of high-level docs for concepts related to single cell models.
  * morphologies
  * labels (region and locset expressions)
  * mechanisms
  * cable cells

It adds some python scripts to the `doc` path, which are called during documentation building to generate images of cell morphologies, regions and locsets that are generated in a separate script that uses the Arbor Python API.

The docs are not complete: the cable cell and mechanism docs are unfinished, and the tutorial section needs to build on these concepts to give more detailed step by step examples.

Rename import method to avoid Python keyword collision.

Fix a Python example that was out of date.

* add parse_swc_allen method

* small fixes and clean up

* add test for incomplete branches and make error messages more consistent

* add extra swc assertion, and remove single sample special case

* Remove spherical root/soma from morphology representations and logic.
* Replace `sample_tree` with `segment_tree`, which represents sample points in pairs and allows the representation of detached branches and cable segments.
* Remove sample point properties and predicates.
* Remove `cv_policy_flag::single_root_cv` as somata with a single attached dendrite will have a representation in the morphology of a single branch.
* Remove `arb::math::{volume/area}_sphere`, which are no longer being used.
* Remove spherical root special cases from `embed_pwlin` and `place_pwlin` objects.
* Add textual representation of `mnpos` in morphology exception text.
* Rename `sample_tree_from_swc` to `segement_tree_from_swc` and remove any special casing for one point somata.
* Remove `msample` and `mbranch` types.
* Update unit tests and python wrapper to reflect the changes.

* Add import method for mechanism catalogues, allowing mechanisms from a second catalogue to be incorporated with a prefix, together with Python interface.
* Expose `global_allen_catalogue`.
* Extend unit tests to cover mechanism catalogue import and check for name collision error cases.

* Move `cv_policy` classes to new source files.
* Remove `cv_policy_flag::single_root_cv`.
* Allow all supplied `cv_policy` makers to be restricted a domain specified by a region: the policy is then applied per component, with the boundary points of the completion of the components included.
* Expose `cv_policy` domain in public interface.
* Add `components` function for splitting an extent into disconnected components. Add comment describing what is meant by connected in this context.
* Rename `super` region function to `complete`.
* Add region expression functions `complement` and `difference`.
* Add new locset expression function `boundary` which returns the extrmal points of each component of a region.
* Add new locset expression `cboundary` which returns the extremal points of the completion of each component of the region, which is what we need to close off CVs generated by region-restricted CV policies.
* Add locset expression `supp...

Update core library:
* support derivation of named mechanisms in catalogue from a string description of the parent.
* fix bug in radois_ge region expression that was finding region_lt.
* fix test for radius_ge region expression

Add support for mechanism, global properties and catalogue shenanigans to python wrapper