generalize applyStatisticalTests

[jakobbossek]

Generalize. Basically we split a data.frame by a categorical variable (here prob), split again pairwise by another categorical variable (here algorithm) and apply tests to one or more numeric columns (e.g., HV, ONVG). So basically, we can write a function dfTest = function(df, by.col, pair.col, val.cols)
This would enable to apply the function not only to result of computeIndicators, but also for results of e.g. computeDominanceRanking.

Example:

ranks = computeDominanceRanking(df)
test.ranks = dfTest(ranks, by.col = "prob", pair.col = "algorithm", val.cols = "rank")

...

inds = computeIndicators(df, unary.inds = list(list(fun = emoaIndHV), list(fun = emoaIndONVG)))
test.inds = dfTest(inds, by.col = "prob", pair.col = "algorithm", val.cols = c("HV", "ONVG"))