Comaring control vs control

[alecdavi]

When running asymptotic_wilcoxon in an OVO manner, results include comparisons of the control group with itself, which is a little confusing for result readability. Maybe it would be better to make sure we only compare groups that are different with the control group, to avoid having something as shown in this screenshot.

[remydubois]

Hello,

Thanks for bringing this up. This is a very valid remark.
I need to do a little bit of twisting in the results reporting because the results object (the pd.DataFrame) is actually the biggest memory allocation, so all manipulations of it should be done with a little bit of care.

Also, this will save a little bit of runtime because fold-change does not have to be computed for this group.

Something I will definitely look after for the next versions.