With much more of the HSC focal plane being covered by bad amps than we had a few years ago, it's much easier for these regions to cause a lot of INEXACT_PSF flagging on the coadd. Given that they're all nicely rectangular, I imagine we could avoid this by adding them to the ValidPolygon regions instead, since CoaddPsf (etc) pays attention to those.
I suspect this isn't super easy, because it'd require tracking the bad amps via something other than the existing defect system, and I think that means a new kind of calibration product. Paul Price, do you have any thoughts on that?
(Credit for this idea goes to Yusra AlSayyad - I'm just making sure we don't forget about it).