Currently, CalibrateTask uses the default LoadIndexedReferenceObjectTask dataset name, "cal_ref_cat", for both the astrometry and photometry reference catalogs. This means that each obs package must configure its reference catalogs individually: there is no reference catalog named "cal_ref_cat". This results in a lot of duplicated config blocks across the obs packages that slowly drift in and out of sync as our reference catalogs are updated, and results in user confusion because they don't know what a "cal_ref_cat" is.

We now have two good quality reference catalogs that should be acceptable defaults for most people wanting to process data with the stack: gaia dr2 and ps1_pv3.

I propose that we use CalibrateConfig.setDefaults() to set these defaults, and remove such settings from the obs packages, except for any specific filterMaps they may need for PS1 (e.g. `i2`->`i`):

astromRefObjLoader.ref_dataset_name = "gaia_dr2_20200414"

astromRefObjLoader.anyFilterMapsToThis = "phot_g_mean"

photoRefObjLoader.ref_dataset_name = "ps1_pv3_3pi_20170110"

I am aware that PS1 is not ideal for LSST operations because of depth and sky coverage, but nothing better currently exists. When a better photometry reference catalog becomes available (e.g. when Gaia releases their coarse spectra--assuming we have a way to use it), we can update the photometry default to better reflect LSSTCam processing.

Any user that doesn't have copies of the above two catalogs can override with their own values: the only change for such a user is that the error message they receive with the default values will refer to an actual existing refcat (potentially prompting them to search for where to get a copy), instead of one that does not exist anywhere.

HSC can keep their config overrides (PS1 for both) after this is implemented while they explore how Gaia DR2 changes their astrometric calibration.