Fix to allow RRTMG longwave radiation (ra_lw_physics=4) to be used with vertical nesting

[rsarthur]

TYPE: Enhancement

KEYWORDS: vertical nesting, RRTMG, longwave radiation, ra_lw_physics

SOURCE: Robert Arthur (LLNL)

DESCRIPTION OF CHANGES: When the RRTMG longwave radiation scheme (ra_lw_physics=4) was
used in combination with vertical nesting, the longwave radiation was incorrect because the
variable NLAYERS was set incorrectly. To fix this, NLAYERS is recalculated based on the updated
number of vertical levels. This fix follows what is done already in the code for ra_lw_physics=1.

LIST OF MODIFIED FILES:
phys/module_radiation_driver.F
share/module_check_a_mundo.F

TESTS CONDUCTED:

1. Regression test runs and gives bit-wise identical parallel solutions with RRTMG and vertical
   nesting (21VN), and is also able to run RRTMG without vertical nesting (case=tropical,
   case=conus):

SUCCESS_RUN_WRF_d01_em_real_32_em_real_21VN vs SUCCESS_RUN_WRF_d01_em_real_33_em_real_21VN status = 0
SUCCESS_RUN_WRF_d01_em_real_32_em_real_21VN vs SUCCESS_RUN_WRF_d01_em_real_34_em_real_21VN status = 0
SUCCESS_RUN_WRF_d01_em_real_32_em_real_conus vs SUCCESS_RUN_WRF_d01_em_real_33_em_real_conus status = 0
SUCCESS_RUN_WRF_d01_em_real_32_em_real_conus vs SUCCESS_RUN_WRF_d01_em_real_34_em_real_conus status = 0
SUCCESS_RUN_WRF_d01_em_real_32_em_real_tropical vs SUCCESS_RUN_WRF_d01_em_real_33_em_real_tropical status = 0
SUCCESS_RUN_WRF_d01_em_real_32_em_real_tropical vs SUCCESS_RUN_WRF_d01_em_real_34_em_real_tropical status = 0

2. A nested case was run with 4 setups, with ra_sw/lw_physics = 1 and 4, and vertical nesting
   on and off on the finest domain (see plot). The non-vertically nested case has 88 vertical levels
   and the nested case has 351. While there are differences in the longwave radiation (GLW)
   depending on whether vertical nesting is used or not, the differences when using option 4 are
   comparable (and slightly smaller) than those with option 1.

RELEASE NOTE: The vertical refinement option now works with the RRTMG radiation scheme.

[davegill]

@rsarthur
We would need this change to use the develop branch as the base branch, as this is considered new development work. It seemed to work OK for me to just swap the base branch, so we are OK.

[davegill]

@weiwangncar @dudhia @smileMchen
Folks,
The proposed mods:

1. are quite small
2. touch only a couple of files
3. require no communications
4. give reasonable results

We should do a 1-month test with these mods to validate that no longer-term biases are being introduced. That would require a set up something like:
test 1 = RRTM / Dudhia + nested
test 2 = RRTM / Dudhia + vertically nested
test 3 = RRTMG + nested
test 4 = RRTMG + vertically nested

[rsarthur]

@davegill
Thanks for looking into this. Attached is my namelist for the radiation option 4 case with vertical nesting.
namelist.input.txt