Problem with test data variables OR with variable_tags

[rod-glover]

At present, we have the following two things:

1. Function variable_tags detects climatological variables based on the following criterion: net_var_name ~ 'Climatology'.
2. Test dataset data/crmp_subset_data.sql defines a large number of variables which meet one or both of the following conditions:
   • long_description includes the phrase "Climatological mean"
   • tests (which I don't trust even in their original form) embodied in pdp_util, classify them as climatological

Item (1) was the result of examining all variables currently in CRMP and observing that their names, sensibly enough, met that criterion.

Item (2) is legacy.

We should fix this. Given that (1) seems to be a functional way to do things for now, at least, and (2) hails very far back, I suggest that we fix the data.