Bug fix for Noah-MP snow, vegetation and urban (#1929)

TYPE: bug fix

KEYWORDS: Noah-MP, snow combine, vegetation fraction scaling, urban ground heat flux

SOURCE: Cenlin He (NCAR/RAL)

DESCRIPTION OF CHANGES:
There are a few bug fixes for Noah-MP related processes:
(1) the snow layer index in snow COMBINE module was wrong, causing model crash when snow layer changes by more than 1 layer within one timestep; This has been fixed by using the correct layer index. 
(2) the ground heat flux sign in Noah-MP column model (positive: downward) is inconsistent with that in urban canopy model (positive: upward), leading to wrong diagnostic grid-mean ground heat flux calculation for urban grid. This only affects the diagnostic value for output. See this issue: #1921 and NCAR/hrldas#114
(3) there is a bug in vegetation fraction (FVEG) scaling for stomata resistance calculation (NCAR/noahmp#92) and canopy interception calculation (NCAR/noahmp#91).

LIST OF MODIFIED FILES: list of changed files (use `git diff --name-status master` to get formatted list):
phys/noahmp/src/module_sf_noahmplsm.F
phys/noahmp/src/module_sf_noahmpdrv.F

TESTS CONDUCTED: 
1. Tested successfully in NCAR Cheyenne HPC for 13-km run over the entire CONUS region
2. The Jenkins tests are all passing.

RELEASE NOTE:  Noah-MP bug fix for snow combination, vegetation fraction scaling, and urban ground heat flux sign.