I spent the better part of a day troubleshooting this with John Parejko, Krzysztof Findeisen, and a few others who chimed in.

We realized the ap_verify_hits2015 dataset has the correct refcat shards, and we "just" need to export them using the gen3 butler and then import them into the repo I plan to use. To this end, we have starter scripts in `/project/mrawls/hits2014-3` called `export_refcats_to_gen3.py` and `import_refcats.py`. The export appears to work, but the import has an error that was discussed on Slack and not yet resolved.

Once the refcats are in place, I have a draft `APTemplate.yaml` pipeline (in the same directory) which should allow me to do single frame processing and then build coadds for use as templates. I can then proceed with DM-27370.

I've managed to export refcats from one repo and import them to another.

The other necessary step (for a pipeline which includes building coadds, like the one I'm running to build templates) is registering a skymap. With all the raws in place, I ran `butler register-skycap . -c name="decam_hits2014_v1"`.

Both the hits2014-3 and hits2015-3 repos now have panstarrs and gaia refcats in them, as well as registered skymaps, for use with decam processing.

The final step I had to do to get my pipeline to run (besides set various configs correctly) was `butler define-visits REPO DECam`.

To use the refcats in calibration (lsst.pipe.tasks.calibrate.CalibrateTask), these configs are necessary:

astromRefObjLoader.ref_dataset_name: 'gaia'

#astromRefObjLoader.anyFilterMapsToThis: 'phot_g_mean'  # not yet implemented in gen3, DM-27843

python: |        

  config.astromRefObjLoader.filterMap = {}        

  config.astromRefObjLoader.filterMap['g'] = 'phot_g_mean'        

  config.astromRefObjLoader.filterMap['r'] = 'phot_g_mean'        

  config.astromRefObjLoader.filterMap['i'] = 'phot_g_mean'        

  config.astromRefObjLoader.filterMap['z'] = 'phot_g_mean'        

  config.astromRefObjLoader.filterMap['y'] = 'phot_g_mean'      

photoRefObjLoader.ref_dataset_name: 'panstarrs'      

photoCal.photoCatName: 'panstarrs'      

connections.astromRefCat: 'gaia'      

connections.photoRefCat: 'panstarrs' 

John Parejko, would you review this? The only code is a functional `APTemplate.yaml` now that I have managed to get refcats for the HiTS fields into my gen3 repo.

One followup regarding skymaps. I mentioned that I ran `butler register-skymap REPO -c name='name'` in both of my gen3 repos. Turns out this resulted in two identical skymaps in all but name, which really confused things when I went to export the deepCoadd templates I made in the 2014 repo to the 2015 repo. With help from Jim, this is how I resolved the situation.

(1) do export with elements=() in the call to }}{{saveDatasets

(2) manually edit data_id mentions of origin repo skymap name to be destination repo skymap name in the yaml created during export (do not edit any file paths!)

(3) skip dimensions skymap, tract, and patch when calling import_.

I will attach all the export/import scripts I used to this ticket for future reference.

A few docs clarifications, otherwise this looks good.