Expose finer grained control over code in update_atmos_radiation_physics?

[spencerkclark]

Over the years we have found it useful to be able to split the time loop of FV3GFS into a finer-grained set of steps. In particular for one application we found it necessary to break the update_atmos_radiation_physics subroutine up into three parts:

• update_atmos_pre_radiation
• update_atmos_radiation
• update_atmos_physics

ai2cm/fv3gfs-fortran#317 illustrates how we proceeded with this in FV3GFS. The update_atmos_radiation_physics subroutine is retained and does exactly what it did before (so this does not change answers and requires no downstream code changes), but is written in a more modular fashion, exposing its steps as additional public subroutines in atmos_model_mod. Would we be open to doing the same in SHiELD?

[spencerkclark]

ai2cm/SHiELD-wrapper#7 illustrates this working in the SHiELD wrapper. Note it points to a branch on my fork of the NOAA-GFDL/atmos_drivers repository; the relevant code changes in atmos_model.F90 are here. I can open a PR if given the green light.

[spencerkclark]

cc: @lharris4 @bensonr sorry for not tagging you earlier (I now see you are not following this repo). I made a draft PR to make the proposed changes a little more visible: #30.