Add a mechanism for performance benchmarking

[msschwartz21]

As mentioned in #59 and #57, it's currently difficult to identify changes that are leading to performance regression. I spoke to a labmate who suggested two possible packages for implementing some form of performance benchmarking:

• https://asv.readthedocs.io/en/stable/
  • Far more sophisticated, but potentially major overkill for what we need and harder to get up an running
• https://pytest-benchmark.readthedocs.io/en/latest/
  • Easier to get started, less sophisticated functionality
  • Github action support: https://github.com/benchmark-action/github-action-benchmark

However there is a major caveat which is that running performance benchmarks as a part of cloud based CI can lead to highly variable results and potential false positives just as a function of what the compute resources were at the time of benchmarking.

My suggestion would be to implement a pytest-benchmark routine that we run on CI, but also run locally when we are concerned about a particular change leading to performance differences.

I'll probably take a stab at implementing something next week when I work with Caroline. Let me know if you have any thoughts on this approach, esp @DragaDoncila @bentaculum

[msschwartz21]

Closed by #62 and #64