Fix method to discriminate climatological variables

[rod-glover]

Resolves #50

Notes for reviewers:

• In tests, the expected values have been changed to reflect the following things:
  • The test data loaded from PyCDS has changed. There are fewer observations (I think); this accounts for some stations with null values for min/max obs time, variable tags, etc. Variable definitions have also been changed.
  • The classification (climo/obs) of variables may have changed.
  • These changes have been verified with independently formulated queries that are shown in tests/test_counts.py::test_database_contents.
• Black has been applied to the files touched in this PR. There are a few misc changes associated with that, not many.

[corviday]

The test data loaded from PyCDS has changed. There are fewer observations (I think); this accounts for some stations with null values for min/max obs time, variable tags, etc. Variable definitions have also been changed.

I saw something like this updating tests for the station data portal backend. Previously the test dataset had explicitly inserted some rows into the VarsPerHistory when it was a managed matview; when it was converted into a native matview the test harness couldn't insert rows, and none were generated "naturally", as that variable didn't actuallly have any associated observations.