Standardize Coadd and Differencing datasets

Ok. I completely agree with removing raw_skyTile, and while I'm a little nervous about dropping raw_visit from HSC in particular, with both your research and K-T's recommendation behind it I'm okay with it.

Merged, with individual paf modifications moved to RFC-232-237

@Jim Bosch I chatted with K-T about raw_visit and raw_skyTile. He says that raw_skyTile is completely obsolete, so I removed all raw_skyTile entries from the obs_base datasets.

I also looked over registry.sqlite3 files for all of the mappers, and none has a raw_visit file which has columns which are not in raw and which are used by any datasets. Since at least in theory, raw_visit is derived from raw, I think it is safe to remove/replace raw_visit with raw.

I have added these changes since the review. Let me know if you approve.

Also, any changes which involve hand editing individual paf files will be moved to the RFC-232-237, that is, the RFC for the individual mapper. I expect that all to be done next week. I've added an issue for those changes. This ticket is specifically to cover global changes involving obs_base.

Would you like to have obs_base contain the template in the former case, so that an override is only needed if the template differs from what is in obs_base?

Yes.

In general, obs_base will have whatever HscMapper has.

Agreed. If you do run into any cases where all of the other mappers agree with each other but disagree with HscMapper, feel free to make HscMapper the one that has to override the default.

@Jim Bosch Sorry if that last comment was not clear. I think that some of the coadd and difference datasets only contain filter, tract, and patch, while others contain camera dependent columns. Would you like to have obs_base contain the template in the former case, so that an override is only needed if the template differs from what is in obs_base?

In general, obs_base will have whatever HscMapper has.
.