Investigate performance regression following #1694

[braingram]

#1694 enabled support for the new skycell reference file. The last regtest for that PR showed performance regressions for a few tests:

[braingram]

@stscijgbot-rstdms

[stscijgbot-rstdms]

This issue is tracked on JIRA as RCAL-1077.

[schlafly]

Thanks Brett. I was worried that something was blowing up in the source cataloging step. Here's the L3 catalog from the regression tests:

The normal catalog looks fine to me, but the multiband catalog has gone a little crazy (spurious green detections in lower right). One issue there is going to be that we should be adjusting the size of the PSF now that we're resampling on a smaller grid, but I'm not sure that that's all of what's going on here. I can't figure out where that's coming from; it doesn't look related to the geometry of the input images.

@larrybradley , something to look at for the next build?

[braingram]

@zacharyburnett mentioned that the increased memory usage may be due to the new size of the table for the skycell reference file.

[larrybradley]

@schlafly Yes, I'll investigate.