Scorecard selector dependencies

[mollybsmith-noaa]

When I was doing the selectors for the MATS scorecards, I made the decision to have the metadata for region, threshold, etc depend only on data-source, rather than jointly on data-source and validation-data-source. There are two reasons why I did this:

1. It makes the code MUCH simpler and less prone to errors.
2. It's not needed anyway. The models that people choose for their validation source are either going to be:
   a) Another version of the experimental model (e.g. comparing one RRFS version to another), in which they'll already have the same regions, thresholds, etc by virtue of our ingest process.
   b) An operational model, which are always ingested on the maximum list of valid regions, thresholds, etc, in which case they'll have more options than the experimental data source, making the data-source selector the limiting factor.

We can always change this setup in the future if we need to, but Jeff and Randy both agree with my current logic.