Benchmarking actor execution

I'm in the process of adding benchmarks to my Actor library, "Stage". Stage aims to run with any type of executor including Crossbeam/Executors and Tokio.

So far, I've found Crossbeam/Executors to be almost three times faster than Tokio. I'd love for some peer review in case I'm doing something silly as I'm a little surprised by the difference. My goal is to have these benchmarks reflect real-world usage of actors; in particular, sending messages between them and creating other actors.

Here the draft PR: Introducing benchmarks by huntc · Pull Request #9 · titanclass/stage · GitHub. Thanks for any feedback!