Benchmarking in CI

[glwagner]

I think it would be wise to set up CI benchmarking. We want to track performance over time. We also want to make sure that new features like those on #236 and #248 don't slow down LES substantially. We can manually benchmark, but having an automated system would be useful.

For this I think we should first create a new directory / environment in benchmarks/ with a small library that allows us to quickly build informative benchmarkable simulations.

We also need CI runners that are suitable for benchmarking. I am not sure github actions meets this. @giordano @bischtob