Observational data checks for diagnostics

In discussions with observational data set providers, we realized that some of the observational data sets have constraints on transformations such as regridding. The data set providers would like to understand how the data is used in the diagnostic for scientific accuracy purposes.
Thi issue is to determine the best way to connect the diagnostic developer (for every diagnostic in the REF) with the obs dataset provider used in the diagnostic and check off if the obs data provider's recommendations for how to use the data for evaluation have been met in the diagnostic.

I suggest having a table with a column marked `checked` or such but other options welcome. Thanks!

Tagging some of the people who might be involved in this but can add more : @nocollier, @dhegedus99 , @lewisjared.