At the moment the KS testing package implemented in the similarity module is easy to pass if the sample of observational data is small (which it often is). We might want to see whether resampling gives different results than the current implementation of the univariate KS testing in xks.