PhotoCalib is now defined in terms of nanojansky, but our reference catalogs are saved in Jy, and remain in Jy when the user loads them. We need to get our refcat loaders to produce nJy to keep our calibrations self-consistent.
It may be enough to just multiply fluxes and flux errors by 1e9 after the data is read from disk, but Simon Krughoff may have other suggestions.
We will have to add a few work-arounds for this to PhotoCal and other tasks that use the Calib object, as it will remained defined in Jy. If only we could do DM-10153 simultaneously (but I think that would be maddening).