Dartboard: give train splits have short baseline data

[jeffjennings]

• Alters crossval.DartboardSplitGridded to give every train split the cells belonging to the smallest baseline bin in the dartboard. Withholds these cells from all test sets.

  • This resolves an issue in which the splitting can result in one k-fold performing quite different from the others (having an outlier loss function and CV score) because the split for that k-fold happened to put some of the smallest baseline cells in the test set. Thus it seems even missing some of the shortest baseline points can significantly affect the modeling.

• Sets the default Dartboard in CrossValidate to set q cells based on the max baseline in the data, not the max of the Fourier grid (which can be quite a bit larger). This ensures the points that every train cell gets aren't too large a fraction of the total vis distribution. I've set the q bins to be the same as in vis_histogram_fig so that the figure made by that corresponds to the splits in a cross-val loop by default (also added args to pass in q and phi bins to CrossValidate).

• Updates the figure produced by plot.split_diagnostics_fig to look a bit better

[jeffjennings]

Example of a split with the new implementation -- none of the splits have red points at the smallest (u,v) locations (but they do still have points at >530 klambda, i.e., including where the weighted visibility distribution peaks):

This figure omits the data (missing in the circle around the origin) that are being given to every train set:

Weighted visibility distribution for reference. The train sets are all getting the cells in the smallest dartboard bin (0 - 530 klambda in this case):

Example of the loss evolution, now without an outlier k-fold: