Redefine fine-res budget for new ShiELD

[nbren12]

The new simulation had different physics options and therefore different diagnostics outputs. Relevant changes included:

• no gfdl microphysics output. Now microphysics is in the physics tendency, and the fv_sat_adj dynamical core phase changes
• New formulation of pressure velocity "omega" resolving this error: Potentially incorrect computation of "lagrangian_tendency_of_air_pressure" NOAA-GFDL/GFDL_atmos_cubed_sphere#40. The correct pressure velocity and corresponding eddy flux are now named "vulcan_omega" and "eddy_flux_vulcan_omega_sphum" for example.

Changes introduced by this commit:

• Adjust hardcoded budget terms to accomodate the changes above. The team suggested that I not overgeneralize this code since won't have a new dataset anytime soon.
• Update test schema to reflect the new training data
• Add unit and long_name information to the outputs
• Simplify the variable renaming logic. Previously "_coarse" was automatically stripped from all the names of the diagnostic output, which made it tricky to know infer what variables were available at a particular step of the pipeline. Now most of the renaming is concentrated into one function.
• Adjust FineResolutionSources to reflect this schema change.

VCMML-310

Closes #347

[nbren12]

Thanks @brianhenn for the unit tests of the mapper. It's sooo much easier working with a code base that has some unit test coverage.

[spencerkclark]

Thanks @nbren12 -- this looks great to me. I agree it's nice to keep the variable names directly read in from the diagnostics constant (rather than removing the "_coarse" suffix).

Note this should close #347.