Recent changes from the HSC side (
DM-4768) were implemented in a hurry, and break CModel forced measurement when the reference WCS is different from the measurement WCS (as is the case with forced measurement on CCDs). This was considered an acceptable temporarily, since forced CCD measurement is currently severely limited by our lack of deblending, but we'll need to fix it eventually.
The fix is trivial from an algorithmic standpoint but may require a bit of refactoring (at least changing some function signatures; maybe more).
This should include re-enabling the different-WCS complexity in testCModelPlugins.py,