Updates to the MYNN surface-layer scheme

[joeolson42]

MYNN surface-layer scheme update:

1. Updates from RRFS_dev, which include updates made for RRFS "Base E" (negligible impact on forecasts beyond the first 10 or so timesteps) + some addition mods below:
2. Reintroduce u* averaging over water, which can help reduce the noisy 10-m winds in the hurricane regime. This has little or no impact in any other regime.
3. Clean-up of the model constants. There was actually a benign bug found with a constant coming in from the interface AND the "use phys_constants". Also, g was defined globally and in some subroutines. This can be further cleaned up later, but this takes out all known issues.
4. Changed "kind=kind_phys" to just "kind_phys", which was requested in a previous PR.
5. Updated some comments that were out of date.

Companion PRs:
ufs-community/ufs-weather-model#1678
NOAA-EMC/fv3atm#640

[BrianCurtis-NOAA]

@grantfirl You may want to remove @ChunxiZhang-NOAA as a code owner.

[dustinswales]

@grantfirl You may want to remove @ChunxiZhang-NOAA as a code owner.

@joeolson42 @grantfirl I opened a PR into this branch replacing Chunxi @Qingfu-Liu.

[grantfirl]

Thanks @dustinswales. Joe is testing this PR individually before we try to combine with #49. I think the latest status is that both individual PRs (this one and #49) are debugging a couple unexpected RT failures, which will be much easier to debug independently.

[grantfirl]

This looks fine to me, assuming RT results are as-expected. As mentioned, there is definitely still room for constant cleanup in a later PR, or when @joeolson42 does a more thorough re-write as discussed offline.

[joeolson42]

All tests successfully compiled and ran to completion. All tests using the MYNN surface-layer scheme were expected to fail the checks due to changes to the solution. The log files are attached.

RegressionTests_hera.gnu.log

RegressionTests_hera.intel.log