Following on PREOPS-894, I noticed that the psf quality statistics (namely scatterSize, scaledScatterSize, and related ellipticity stats) that are currently only computed in PsfWcsSelectImagesTask would be more useful if they were computed in ComputeExposureSummaryStatsTask. This move would have the following advantages:
1) We would have access to these useful quantities for all exposures, not just those that fail the test (via logging).
2) We would have access to these useful quantities even when PsfWcsSelectImagesTask isn't configured to run.
3) We can remove the sourceTable input from MakeWarpTask which would speed up this task when PsfWcsSelectImagesTask is configured to run.
On this ticket I will then move these quantities (and a few others, particularly the number of psf stars used for the psf model) into the visit summary table.
Note that originally I hoped that this could be part of the MeasurePsfTask but it needs to be computed in calibrate rather than characterize so that we have the necessary psf models and star sizes computed.