Code review comment for lp:~maddevelopers/mg5amcnlo/virtual_polyfit_EW

Thanks Rikkert,
the code changes look correct, for what I can understand of the polyfit module
I have tried p p > t t~ [QCD QED], results are compatible within the uncertainty (also split-order by split-order). The Polyfit run is generally (much) faster:
If I require 1 per mil accuracy for that process, without polyfit it took 18 minutes on my macbook, while with polyfit less than 8, so great job!!