call to populate surf_diff for implicit coupling

[JosephMouallem]

surf_diff is moved to the physics to be consistent with the AMx logic.
Call to populate surf_diff variables after the downward sweep.
Depends on:
NOAA-GFDL/GFDL_atmos_cubed_sphere#408
NOAA-GFDL/SHiELD_physics#83

[lharris4]

So if I understand, this set of PRs is to follow the logic of AMx and to encapsulate the required data types and movement within the appropriate components:

• surf_diff_type is encapsulated within the SHiELD physics
• populate_surf_diff() is now within the SHiELD driver in FV3

I see that populate_surf_diff() is now quite a bit more comprehensive in its calculations than the code it replaced in atmos_driver. Is this to improve the accuracy of the near-surface derivatives used for the land-atmosphere implicit coupling?

[JosephMouallem]

This is true for the most part.
populate_surf_diff actually adds (not replaces) new calculations needed to populate the variables under surf_diff which are needed by the full coupler fast loop for land and ice implicit coupling.