TL/AD code sync with WRF changes

[liujake]

TYPE: WRFPlus sync

KEYWORDS: TL/AD code, Sync

SOURCE: Internal

DESCRIPTION OF CHANGES:

1. Recent changes in WRF lead to WRFPlus compilation failure. This update fixed compilation failure.

2. This commit allows WRFPlus/WRFDA-3DVAR/4DVAR to work properly with namelist setting
   &dynamics
   hybrid_opt=0,
   etac=0.0,
   use_theta_m=0,
   /
   , which means MOIST_THETA-T0=T for WRFDA's input and output files and hybrid vertical coordinate is reduced to the original mass coordinate with the values of 8 1D variables in the file:
   C1F = 1
   C2F = 0
   C3F = znw
   C4F = 0

C1H = 1
C2H = 0
C3H = znu
C4H = 0

LIST OF MODIFIED FILES:
M Makefile
M Registry/Registry.EM_COMMON.tladj
M Registry/Registry.tladj
M var/da/da_main/da_update_firstguess.inc
M wrftladj/module_first_rk_step_part1_ad.F
M wrftladj/solve_em_ad.F

TESTS CONDUCTED:
[x] No test for WTF model tests (not needed)
[x] Passed 3dvar, wrfplus, and 4dvar in WTF using new wrf file (with hybrid_opt=0, etac=0.0, use_theta_m=0) with gnu compiler
[x] Run through all WRFDA regression tests (including two DA/forecast cycling cases) with gnu and ifort (with modified wrfinput/fg input files).

[liujake]

can someone approve this?

[davegill]

Approved

[liujake]

@davegill @weiwangncar @dudhia
Finally I am able to run through all WRFDA regression tests with the latest code. I can not get 'match'
(WRFDA regtest compares new output to benchmark output, I see some small diffs mostly for MOIST_THETA-T0, which I believe it is expected. New output from the 'new' WRFDA regtest will become new benchmark output for the future regtest). Please take a look at my revised PR message and see if I shall push them in for friendly release #2. if it is in, you will no longer need 2 separate tar files for WRF and WRFDA.