[WIP] Setup benchmark suite

[timkpaine]

xref: #38
Following @trallard's suggestion, exploring using asv to build a benchmark suite, starting with @AdamGlustein's code in #322 (comment).

Example output, running on just my machine through a few commits:

[timkpaine]

get some standardized machines before putting in raw data

[ptomecek]

I had missed the original discussion at #322 , that explains where a lot of this came from. Would be good to tidy up in any case. There are a whole bunch of benchmark tests that I want to write after...

[timkpaine]

Example from other PR:

[AdamGlustein]

I had missed the original discussion at #322 , that explains where a lot of this came from. Would be good to tidy up in any case. There are a whole bunch of benchmark tests that I want to write after...

That benchmark script was just something I hacked together, I didn't intend for it to be a canonical example.
If we are making standard benchmarks then we should use a more general framework.

[timkpaine]

If we are making standard benchmarks then we should use a more general framework.

I did a first pass approximation of a more general framework, extnded to the quantile function to demonstrate https://github.com/Point72/csp/blob/fb36b1b6f448d8c39aaab4edfb4b0ba7eed82ed7/csp/benchmarks/stats/basic.py

[timkpaine]

@arhamchopra would be good to see if we can use our benchmark infrastructure to run csp benchmarks as well, and maybe fix some of the ugliness I introduced in this PR if we can.