Implementation of MYNN-EDMF and MYNN-SFC submodules

[joeolson42]

This PR adds the official MYNN-EDMF and MYNN-SFC submodules to MPAS with via manage_externals. Some notes on the implementation are below:

1. The submodules were added simultaneously due to the common package (bl_mynnedmf_in) used for both schemes (following the previous implementations).
2. Two additional packages are added to control the allocation of 10 optional 3d output arrays and 5 tke budget arrays.
3. a new cloudiness option is added to handle the processing of the MYNN-EDMF subgrid clouds analogous to radiation driver in WRF.
4. The implementation was complicated by the earlier implementation of MYNN schemes, so renaming of key modules and subroutines was required to allow both submodules to work as alternatives to the other versions. The other schemes should still work but will hopefully be removed.

[mgduda]

When this PR branch is merged to the HEAD of the develop branch, I'm seeing a compilation failure with the GNU compilers:

mpas_atmphys_driver_pbl.F:1445:39:

 1445 |                   bl_mynn_stfunc     = bl_mynn_stfunc      ,                                              &
      |                                       1
Error: Keyword argument 'bl_mynn_stfunc' at (1) is not in the procedure; did you mean 'bl_mynn_output'?

[joeolson42]

bizarre... I must not have done a "git add mpas_atmphys_driver_pbl.F" before the commit. That line is not in my code. Will update shortly.

[mgduda]

Thanks for the update. All looks good compilation-wise with the changes in 5a2cdc6.

[mgduda]

Declaring icloud_bl and spp_pbl here as local variables overrides the variables of the same name imported from the mpas_atmphys_vars module. Since these local variables are pointers, and since they're not set by a call to mpas_pool_get_config when using config_pbl_scheme = 'bl_mynn', we'll get a segmentation fault (the apparent location of the segfault may depend on compiler).

One option to avoid this issues might be to choose different names for the local pointer variables icloud_bl and spp_pbl, and to use those different names only in MYNN-EDMF cases (leaving behavior with the MYNN scheme unchanged).

[joeolson42]

I think it's best to remove icloud_bl from the pbl driver entirely (at least for "bl_mynnedmf"). icloud_bl really only has meaning in WRF (package declaration and coupling to radiation). In MPAS, it has no functionality, even for the cloudiness driver. I'll update the MYNN-EDMF submodule, the WRF registry and pbl driver, and this PR, to remove (or lessen) the need for icloud_bl. As for spp_pbl, the spp development is almost finished, so it will be available soon and will not be hardwired in the mpas_atmphys_vars module inthe near future. For now, I can remove that line as well.

[joeolson42]

OK, those lines have been removed, but I'm not sure it's improved. Those "options" are still hardcoded in the mpas_atmphys_vars module, but that was the original design. The other MYNN scheme should still work.