update to MYNN-EDMF, with submodule pointers for fv3atm and ccpp/physics

[joeolson42]

Overview of changes:

• This commit should be considered step one of the "universalization" process to generalize module_bl_mynn.F90 both for all dycores. Much of this is strictly reformatting or repositioning the driver to the top of the module - a request from collaborators, but some of it involve removing ifdefs, and the handling of the model constants. More commits to complete this work will follow. Despite the huge number of lines changed from this work alone, there is no change in behavior/performance.
• fixed some old crash cases (improved numerical stability)
• conservation work - removes all negative qv printouts
• A bug fix for the interpolation of the mass-flux clouds from interface levels to mid-levels which cause excessive LWP in shallow-cumulus fields
• Removes a few unused variables/arrays
• Fixes for the smoke/chemical mixing (works but still not in final state - another commit will follow)
• Some tuning work to improve the excessive cloud bias in polar regions.
• Change the default mixing length option to 1.
• Adds option to output the TKE budget (optional additional output - does not change behavior).
• General clean up of code and updates to many comments (also responsible for many lines of code changes).

Testing

A combination of case studies and retro tests using the intel compiler. The regressions tests used both intel and gnu. Most testing was performed on Hera.

• hera.intel
• hera.gnu
• orion.intel
• cheyenne.intel
• cheyenne.gnu
• gaea.intel
• jet.intel
• wcoss_cray
• wcoss_dell_p3
• opnReqTest for newly added/changed feature
• CI

Dependencies

NOAA-GSL/fv3atm#121
NOAA-GSL/ccpp-physics#121

[SamuelTrahanNOAA]

I think you meant to point your "Dependencies" section at these:

NOAA-GSL/fv3atm#121
NOAA-GSL/ccpp-physics#121

[joeolson42]

Yes, thanks

[tanyasmirnova]

Approved.