Wrong units for several LSM-related variables

[grantfirl]

The following LSM-related variables appear to have the wrong units in the CCPP metadata:

• soil_moisture_content (listed as kg m-2, but is actually m when used in the CCPP; i.e. it has not been multiplied by the density of water yet)
• total_runoff (same situation as soil_moisture_content)
• surface_runoff (same situation as soil_moisture_content)
• subsurface_runoff_flux (listed as g m-2 s-1, but are actually in mm s-1, or kg m-2 s-1 equivalent after conversion to m and multiplication by the density of water)
• surface_runoff_flux (same situation as subsurface_runoff_flux)

The first 3 variables are output in FV3 from GFS_diagnostics.F90 and use a "conversion factor" of 1000 (which I'm interpreting as the density of water in kg m-3, but one could interpret it as just a number, but either way it affects the units).

There are at least 2 solutions for the first 3 variables:

1. modify the units within the CCPP metadata to reflect their actual units (not those that are output after a conversion factor)
2. remove the conversion factor in GFS_diagnostics

Solution 1 would necessitate using non-canonical units for the quantities, but would only require changes to the metadata. Solution 2 would require a change to the FV3 source code but the CCPP could then use the expected, canonical units for these variables (preferred).

For the runoff fluxes, only a metadata change should take place for Noah and NoahMP. The RUC LSM will need to convert to the new units since it had been coded to use the "wrong" units in the CCPP.

[climbfuji]

This was fixed with #307 and #308.