Add py-jupyter-server-proxy to spack.yaml

This package enables users to start web servers in JupyterLab and access them, which is useful in rare edge cases like post-processing of viztracer results.
The latest spack package installs py-jupyter-server-proxy v3.2.2.
The current version on pypi is v4.4.0.
https://packages.spack.io/package.html?name=py-jupyter-server-proxy
https://jupyter-server-proxy.readthedocs.io/en/latest/
https://github.com/jupyterhub/jupyter-server-proxy
https://pypi.org/project/jupyter-server-proxy/

packages/py-snudda/package.py

@@ -46,11 +46,15 @@ class PySnudda(PythonPackage):
     depends_on("open3d+python@0.18:",   type=("build","run"), when="@2:")
     depends_on("py-quantities@0.16.2:", type=("build","run"), when="@2.1.14:")
 
-    # snudda tarballs in pypi do not include the tests/ dir: just use default spack tests for now
     # @run_after('install')
     # @on_package_attributes(run_tests=True)
     # def install_test(self):
-    #     python('-m', 'unittest', 'discover', '-v', '-s' './tests')
+    #     with working_dir("tests"):
+    #         from pathlib import Path
+    #         excluded_tests = ['test_sonata.py', 'test_snudda_path.py', 'test_frequency']
+    #         test_files = [str(p) for p in Path(".").rglob('test_*.py') if p.name not in excluded_tests]
+    #         if test_files:
+    #             python('-m', 'unittest', '-v', *test_files)
 
     # blender is for now an optional dependency: leave out of import_modules to avoid errors in tests
     skip_modules = ['snudda.plotting.Blender', 'snudda.plotting.Blender.io_mesh_swc', 'snudda.plotting.Blender.visualisation']