Can we add benchmarking?

[JSAbrahams]

Summary of current issue

Currently, we perform no benchmarking.
In general it would be nice to as the project matures include some sort of benchmarking to check to what extent there is regression in performance.

Proposed change

• Add a benchmarking library to the project, non -intrusive
• Add a bot perhaps which comments on each PR how performance changed.
  Similar to codecov, so that we can see changes in performance without regressions being blocking.

[JSAbrahams]

Perhaps https://github.com/bheisler/criterion.rs is interesting