Fill out grid DDT from MPAS

[grantfirl]

Hey Dustin,

I'm not sure if you want to periodically merge my work back in or what, but here are my first code changes if you do!

I'm going through the timestep_init phase of the RRFS_MPAS suite and found that there are a bunch of uninitialized variables in GFS_phys_time_vary, so that is where I'm getting my feet wet.

This code does the following:

• adds a new call during the atmos_model.F90/atmos_model_init() subroutine to call the newly-created atmos_coupling.F90/ufs_mpas_grid_to_physics() that accepts the newly created GFS_grid_type DDT.
• atmos_coupling.F90/ufs_mpas_grid_to_physics() basically does the same thing as GFS_init.F90/grid_populate(); sets lat/lon in rad/deg, sinlat, coslat, area, and dx.

I'm using areaCell from MPAS for the area, which is straightforward enough. For dx, rather than sqrt(area) as is done for FV3, I found how it is calculated in the standalone MPAS GWD driver, using the config_len_disp MPAS configuration variable or the nominalMinDc, whichever is available. I copied this code from mpas_atm_core.F/atm_core_init() from the standalone.

This compiles/runs.

[grantfirl]

Next step is surface data...

[grantfirl]

I should probably remove the ccpp-physics hash update -- I am using a test branch that isn't meant to be shared.

[dustinswales]

@grantfirl Thanks for these changes.
Gathering changes in my branch is fine for the time being.