Statistical Testing after Clustering

[Cryaaa]

As a last step in the workflow it could be nice to have a widget which based on a clustering can tell you in which features the groups differentiate from eachother. We could allow for the choice of a few statistical tests (although a good default choice could be the Mann–Whitney U test) and a custom alpha to determine significant results. Also showing histograms of the features for which clusters differ significantly could be interesting.