feature request: permutation testing of input sets

[mjsduncan]

given an input set and a parameter file, gnerate models on a dataset with randomly permuted category labels to establish the statistical significance of ensembles generated by moses from the correctly labeled dataset.
ideally the results from previous moses runs could be browsed so the data and parameters from the results to be tested for significance could be selected in the input interface and the data and parameters from the selected previous analysis would be used to automagically run the permutation test and return a summary report of the computed significance of the chosen analysis.