Implement lightweight benchmarks

[hyanwong]

It would be really useful to have some sort of lightweight benchmarking setup so that we can check whether changes to tsdate affect speed and memory usage. I imagine that we wouldn't run this during CI, (or perhaps, only run a minimal version to check the code works, but without the benchmarks).

@benjeffery pointed out that there are benchmarks for tskit at https://github.com/tskit-dev/tskit/tree/main/python/benchmark and we might be able to copy the setup there.

This is currently stopping us fixing e.g. #217