M31 Bulge sources ignored with a hack

[eugene]

Astrometry analysis in the field of M31 was crashing for GPC1 because of the large number of sources in the bulge, potentially in conjunction with the large number of dark-current sources. the significance of matches was washed out by the background and the noise sources matched the bulge to easily. I hacked a solution to ignore the bulge region when loading the ref sources, but a better solution is needed. some options:

• detect and ignore clumps in the input lists (raw and ref). we could use autocorrelation to find clumps and drop them from both.

• have a list of bad regions specified in the recipe to ignore.

• enforce a uniform distribution of sources in the input lists somehow?