enh/166-vs-perf-plot

[jharwell]

For any univariate batch criteria that generates measure(s) OTHER than raw perf
plots, a correlation graph between the measure and the raw performance should be
generated, so that the R^2 value (i.e. how much of the performance is
explained/correlated by the OTHER measure(s)) can be generated. This would
require a new 2D scatterplot graph.

For bivariate criteria this would require a new 3D scatterplot.