Add benchmarks

[saschatimme]

I think it would be valuable to setup a proper benchmark enviroment. With this we could track performance regressions as well as it would be easy to compare different algorithms.

[chriscoey]

Are there any benchmarks against other packages like PHCpack?

Maybe with the recent progress on PARTR in Julia (JuliaLang/julia#31398 etc) there will be more opportunities for speed improvements on the horizon.

[saschatimme]

There are no up to date benchmarks. But it would be nice to have some automatically generated ones which automatically get rendered into a table. For my experience, we usually outperform Bertini by at least an order of magnitude and PHCpack similarly (although there it obciously depends on the discrepancy between mixed volume and total degree). But yes, plain numbers would be good to have.

Regarding PARTR, our plain solve command is already multi-threaded by default so I don't think we get much speed up there. Whereas the monodromy_solve routine is currently not multi-threaded since I wanted to wait for PARTR to land.