XCA.ask gives poor mask, and reverse ordering

[maffettone]

XCA should be returning values along a consistent y in clusters of interest.
It also is doing argsort in the wrong order.

ae-gpcam/companion/xca.py

Line 85 in e0196e1

jdxs = np.argsort(self.dependent[:, phase])[np.abs(self.independent[:, 1] - current_y) < 4.5/2]

[maffettone]

Fixed in eefba97.

main, prby-figure and reconstruction branches left for posterity with December experiment.