benchmark tests

[kapilt]

given large cardinality resource sets, it would be good to evaluate the performance of cel-python against ~1000-~10000 and look for profile based optimization opportunities as well to have a general sense of the relative performance. the type casting me a bit concerned that we may end

[slott56]

I'm going to move some of this to the wiki pages because there's a lot of space for back-and-forth outside the core issue of "get profile data."

See https://github.com/cloud-custodian/cel-python/wiki/Early-Profiling-Data

A future release will have a benches folder with benchmarking and the profiling.