Add single-point transient tests that test against expected subgrid areas

[billsacks]

In order to prevent bugs like #538 in the future: We could have one or more new test types that are meant to be used in a single-point, transient simulation (though we might want to test a non-transient simulation, too): run for some known amount of time (like an SMS test), then compare subgrid areas (PCT_LANDUNIT, PCT_NAT_PFT, PCT_CFT, etc.) against some expected values. These expected values would match the values used in creating this single-point dataset; they can probably simply be hard-coded in the python test script.

This would ensure that, even if there are answer changes in a tag, the expected behavior of subgrid area changes remains correct.

I imagine that this would be implemented by reading in the test's final history file using the python Netcdf4 library and checking the desired variables in python code.

[billsacks]

@ekluzek wondered if we would need some changes in CIME to go along with this. I don't think we do: I think we can start off with a custom test type defined in CTSM that does the necessary comparison and uses the existing COMPARE_XXX phase (where XXX can be anything you want) to set the results. For example, we could call the result, COMPARE_EXPECTED.

Long-term we might want a generic test type with some common functionality that can be extended for this purpose, but I'd argue for starting with a couple of specific cases before we try to write something general.