Add a modified version of the benchmark cell example that uses dry run scaling.

- Add support for profiling externally loaded mechanisms.
- Move the profiler calls out of the generated C++ code into the `arb::mechanism` methods. 
- Remove the `--profile` flag from the modcc flags. 
- Replace `_` delimiter with `:` in profiler. 
- Replace `const char*` with `const std::string&` for profiler region representation.
- Profiler may now contain empty regions that were registered but not profiled, so add some code to filer those out when generating final profile.

- Bump CMake to 3.18 for better CUDA support.
- Clean-up our use of said support.
- Remove old cruft and workarounds.
- Maintenance in CI: enable MacOS 11; bump CMake versions to 3.18 / 3.22; bump bors CMake to 3.18

Fixes #1790 

- Implement a new constructor for `domain_decomposition` from a vector of `group_descriptions`. The new constructor performs various checks on the given `group_descriptions` and throws an appropriate exception if any problem is found. It is the only available constructor for `domain_decomposition`.
- Make the various members of `domain_decomposition` private and implement their getters.  
- Expose the `domain_decomposition` constructor to the python interface through a `partition_by_group` function.
- Add unit tests. 
- Add documentation.

Documents the release cycle.

Refactor the CV-cable logic from `fvm_layout.cpp`. Specifically, move `cv_geometry_from_ends` into the new public header `cv_data.hpp` and rename it to `cv_data_from_locset`, and move `pw_over_cable` into another new public header `util/piecewise_over_cable.hpp`. 
Create a new public class `cell_cv_data`, similar to the private `cv_geometry` but which only carries the cv information for a single cable-cell. Add a new constructor to `cv_geometry` which takes `cell_cv_data` as an argument. 
Add new public function `intersect_region` that takes as arguments a cable-cell, a region on that cell and a `cell_cv_data` object and returns a vector of CV indices and proportions belonging to the region. 
Add docs, python wrapper and unit tests. 

Fixes #1720. 

Replace the `partition_load_balance` function assumed that gap-junction connections are always double-sided (i.e. if gid `x` has a gap-junction connection from peer gid `y`, then gid `y` must also have a gap-junction connection from peer gid `x`). This used to be a requirement that was checked prior to #1682. Since then, single-sided gap-junctions are in principle allowed, but `partition_load_balance` still operates under the bidirectional gap-junction connection assumption resulting in some gids being present in multiple cell-groups. 
This PR modifies the `partition_load_balance` function to do the following: 
1. On each rank, generate a gj_connection list per cell. (gj_connection list is a vector of gids that have an outgoing connection ending at the cell under consideration). 
2. On each rank, gather all the gj_connection lists for each cell in the network. 
3. On each rank, modify the global list of gj_connections in the network to make all gj_connections bidirectional. 
4. On each rank, use the global list of gj_connections in the network to partition the graph. 

Fixes #1767.

- make fmt a dependency of libmodcc, not modcc
- set FMT_HEADER_ONLY as a compiler flag to avoid double definition

* Create a script that generates `diff.log` if git submodules have tags more recent that the commit they're currently at.
* A Github Action runs this script monthly.
* _If_ updates found, Github Action creates an Issue like so: https://github.com/brenthuisman/arbor/issues/9
* Actually updating the git submodule and Spack file are still manual

First step in addressing #1731

Check for errors in the solvers before the `finalize` stage and return early if errors were encountered. 

Fixes #1722.

* Add documentation for collaboration on a PR using just git
* Add the stage of Issue creation to PR workflow.
* Correct header level in test docs

Add the lmorpho utility back to Arbor

The original utility was removed because it was a maintenance burden with no users.
Now users are requesting the feature, this PR adds the old code, and updates it
to work with the new morphology description API.

* add a test that tests that a catalogue built separately can be loaded via the Python interface
* further simplification of dynamic library support
  * move all platform-specific code into the cpp implementation and out of header.

Fix various non-standards compliant code when compiled with elevated warning levels (-Wall -Wpedantic). Here is a list of the problems addressed:

**preprocessor**
- expansion of empty `__VA_ARGS__` in `ARB_PP_FOREACH`
- named variadic macro arguments in `TRACE` macro

**glibc** 
- deprecated `mallinfo` call for newer glibc versions

**simd**
- non-const C-array sizes

**modcc**
- generation of C-arrays of size 0

**switch/case**
- switch-case ranges

**tuple_foreach**
- compound literals

Use `constraint_category` only when reading/writing data sources indexed by `node_index`. Every other indexed data source assumes that the underlying index belongs to the category `index_constraint::none`. 
Fixes #1734.

Fixes #1753

Extend the `pwlin_place` interface to find the location of a morphology that is closest to a 3d coordinate.
* extend `arb::pwlin_place` interface
* python wrapper
* unit tests
* documentation

Addresses a feature request by @Helveg that I don't think ever had a ticket assigned to it.

Fixes #1661 and #1108.

* Consolidate default_concurrency, default_gpu functionality into arborenv/default_env.hpp, .cpp.
* Add new arborenv function `default_allocation()` that uses environment variables ARBENV_NUM_THREADS and ARBENV_GPU_ID to create an `arb::proc_allocation`, with a fallback to `thread_concurrency()`.
* Use arborenv-specific exceptions instead of generic std::runtime_error etc.
* Update examples, docs to suit.

Fixes #988 (except for reworking all the unit tests!)

tinyopt README.md had an example which was not updated to the 1.0 API. Fixed upstream, and this is a copy.

* Add test that is sensitive to small radius loss of precision in ixa computation.
* Add test that checks for consistent behaviour when there is an (isolated) zero radius point in the morphology.
* Special case p==1 q==1 case in ratelem so that it can interpolate ixa in the presence of non-finite interpolants.
* Change naming in pw_element/pw_elements: 'element' refers to the extent+value pairs comprising a 'pw_elements' object; 'value' refers to the value associated with an element; 'extent' refers to the closed interval which is the support of the element.
* Allow values (but not extents) in a pw_elements object to be mutable, using proxies for iterator access.
* Write piecewise zip (pw_zip) in terms of a lazy pw_zip_view. Rename zip functions.
* Document pw_elements more thoroughly.
* Simplify and document embed_pwlin.cpp routines, expressing everything in terms of simple interpolate/integrate operations.
* Represent integrated inverse cross-sectional area in the embedding by multiple piecewise-rational functions over a branch, each of which contribute separately to an ixa-based integration, so that precision loss associated with small (or zero) radii can be avoided.
* Comment embed_pwlin.cpp more thoroughly.

Fixes #1526 

Add SVE implementations of `sin` and `cos` using `std::sin` and `std::cos` respectively. 
Fixes failing CI on a64fx architectures for vectorized builds.