Actions workflow for profiling unit test runs

It would be useful to have a CI workflow that runs cProfile on the unit test run (or perhaps some custom script that is more useful to profile) for the jobs with their differing matrix variables (currently OS & Python version), that we can run ad-hoc.

It would allow us to pinpoint identify performance differences across platforms and Python versions. In particular I raise this since I have noticed the MacOS jobs all consistently run significantly slower than the Ubuntu jobs over the current test suite. With profiling stats we could investigate that with detailed outputs and without having to rummage through the job logs for timings. Otherwise it is hard to check up on potential Mac-specific bottlenecks, not having a Mac at hand.

Ideally we can have a workflow that grabs the profiling stats for unit test runs with each matrix variable and then runs a final step with some logic to determine any major differences in outputs, to inform us of any significant platform or version-specific differences. Or runs a vis tool like snakeviz or gprof2dot to eyeball for noticeable differences.

No file or test path is named. Start by reviewing the existing GitHub Actions unit-test matrix and its OS and Python-version jobs, then determine how profiling outputs would be collected and compared on demand. Done means an ad-hoc workflow produces per-matrix profiling results and highlights meaningful platform or version differences.