Running measureCoaddSources.py, I'm getting a RuntimeError: Unable to match sources. The patch I'm running on doesn't have a lot of pixels illuminated, so it's not surprising that there would be no matches, but that fact shouldn't cause the operation to fail. The behaviour of the matcher has changed in this respect: before the refactoring of astrometry, AstrometryTask.useKnownWcs would not raise an exception, but log an error and return an empty list of matches (it successfully matched zero sources, which might be cause for concern).

I think we should fix the matcher to return an empty list (as it did previously), but perhaps a case can be made that measureCoaddSources should catch the exception and continue. I think that's wrong because measureCoaddSources would have matches = None rather than an empty list, which communicates something different ("I have no knowledge of matches" vs "I tried to match and there's nothing"); and measureCoaddSources shouldn't simply set matches = [] because it's not its responsibility to guess a type for what its subtask returns (breaks encapsulation).