ASV Benchmarks

[adam2392]

Is your feature request related to a problem? Please describe.
Similar to numpy, scipy and scikit-learn and many other performant libraries, we should implement an ASV benchmarking suite that will then allow us to gauge how to improve the runtime and memory performance of the CI test suite. This will then map to significant improvements downstream in dowhy, dodiscover, etc.

Describe the solution you'd like
Add asv benchmarks. See scikit-learn repo for some boilerplate.

Additional context
We need to figure out where to host the resulting plots and add a CI pipeline to handle this. Not sure exactly on details related to this yet tho...