PhotoCalib is now defined in terms of nanojansky

I just merged DM-16650, which has redefined PhotoCalib to be a conversion from counts->nanojansky. For several reasons, this necessitated breaking backwards compatibility with old persisted PhotoCalibs objects, so if you have run jointcal in the past, attempting to read the older files will raise an exception.

The PhotoCalib API has changed to reflect the new output units: the various overloads of instFluxToMaggies() have become instFluxToNanojansky().

Our reference catalogs are still in Jansky, as is the old Calib object. Updating the output of the reference object loader to be in nJy is the final piece necessary to implement RFC-549, but I have implemented work-arounds in the code to deal with this discrepancy for now. Once DM-17029 is finished (thus finishing RFC-549), I will start on replacing Calib with PhotoCalib (DM-10153).

Big thanks to @erykoff for his work reviewing my math and code.

Related Community discussion and the summarizing Project Science Tech Note that lead to this change.