HWRF SAS deep/shallow convective schemes

[mzhangw]

This branch implements HWRF deep/shal SAS schemes into CCPP. They are using the same codes as GFS SASAS schemes (imfdeepcnv/imfshalcnv=2). To switch to HWRF version, turn on hwrf_samfdeep and hwrf_samfshal in NML:

hwrf_samfdeep=.true.
hwrf_samfshal=.true.
asolfac_deep = 0.89 !#GFS:0.958
asolfac_shal = 0.89 !#GFS:0.958

The associated PR : NCAR/fv3atm#23

[climbfuji]

I know this PR is work in progress, but I would like to raise concerns that I have early on in the process. While the idea of having one unified codebase for the SAS and HWRF SAS convective schemes is good, I am worried about the computational performance (and the readability to a lesser extent). There is a large number of "if" statements inside do loops, and the compiler may not be able to optimize and rearrange all of those. I think this deserves some discussion, especially in moving forward with our physics unification efforts. @ligiabernardet @llpcarson @grantfirl

[mzhangw]

I have modified HWRF sas .i.e. moving hwrf_samf* out of the loop as suggested and run the full RTs. All seems good EXCEPT these two:
fv3_ccpp_stretched_nest_prod
fv3_ccpp_regional_c768_prod

However, the corresponding RTs with REPO passed the RTs and fv3_ccpp_stretched_nest_debug_prod passed RT too!

[climbfuji]

I am closing this PR, since it will be merged to master as part of #423 and to dtc/hwrf-physics as part of #433.