Discussion: Refactor NRemover

[tkchafin]

I wonder if it might make better sense to re-structure all of the Nremover functions as GenotypeData methods? In each case could either modify in-place or return a deep copy of current GenoData object.

This would allow us to be more flexible with the order of operations, for example a user could do GenoData.filterPop(..).filterInd(...) or GenoData.filter(miss=0.5).filter(Mac=2).filterInd(0.5) etc, so you could chain operations similarly to how you might do with a pandas dataframe?

Could include a logging module to track what order operations were done, which filters applied/ data removed at each step, so the sankey plot would still work?