trying out adding benchmarks to github actions CI

[ajfriend]

As worried about in #523.

My GH Actions knowledge is rusty, so more than happy if anyone has suggestions to clean up the test-bench.yml.

[coveralls]

Coverage decreased (-0.2%) to 98.513% when pulling ffde47e on ajfriend:gh_actions_benchmarks into f2786a4 on uber:master.

[nrabinowitz]

Looks good, I can see the benchmark output in the checks. Would be awesome if we had some coveralls-style way of reporting this, but 🤷

[isaacbrodsky]

I wonder if there's some way we can collect and monitor benchmark performance, like https://github.com/marketplace/actions/continuous-benchmark or https://werat.dev/blog/running-benchmarks-for-pull-requests-via-github-actions/?

[isaacbrodsky]

Right now we don't have a way to indicate building just the benchmarks without running them in the same invocation. If the build part needs to be sped up we could turn off some other build options (building tests, etc.) in CMake, or add a "build without run" option for benchmarks.

[ajfriend]

I wonder if there's some way we can collect and monitor benchmark performance, like https://github.com/marketplace/actions/continuous-benchmark or https://werat.dev/blog/running-benchmarks-for-pull-requests-via-github-actions/?

That'd be nice! I'm deferring it to #530 just so I can get this landed, finally 😄