Quality benchmarks

[makspll]

BMS is in dire need of benchmarks, I want to ideally have them run in CI and get published somewhere visible, probably the docs.

I want the benchmarks to give consumers a potential idea of how slow/fast various common operations are such as:

• Reflecting based on depth of reflection
• Calling a dynamic function
  etc.

We should compare against something people will be familar with, for example c# reflection

TODO

• Initial benchmarks infra on bencher.dev
• Expand benchmarks
• Integrate benchmarking into xtask
• Compare PR benchmarks against main every time, notify of significant regressions or improvements
• Add c# benchmark baselines

[makspll]

I am starting work on this in #381