Quartet of bug fixes for: c3 scheme, quilting restart with 32-bit physics, and string length mismatch in dycore plus PR #705 and #699

[SamuelTrahanNOAA]

Description

Bug fixes:

1. The write component no longer calls recover_fields for restart files. (From @DusanJovic-NOAA.) Doing so causes sporadic crashes due to corrupted longitudes when 32-bit physics is used. This only happens for restart files, who don't need recover_fields anyway. Hence, this PR restricts that call to history files.
2. Correct a string length mismatch in the dycore. This causes an abort when the gnu compiler is run with debug flags turned on. One of the checks enabled is to abort on standard violations of this type.
3. Changes some subroutine array argument declarations so the c3 scheme can handle null pointers. This is required by the CCPP coding standards anyway, for that very reason. Without this change, the FV3_HRRR_c3 suite crashes with the gnu compiler.
4. Initialize arrays in dycore. (From @DusanJovic-NOAA.) Two arrays in the dycore were not initialized. That caused crashes in the quilting restart when 32-bit physics is used.

Included in this PR are:

• Remove nine 3D arrays from CLM Lake model #705
• GF radar reflectivity for RRFS realtime runs #699
• Unify the missing value in physics and dynamics NOAA-GFDL/GFDL_atmos_cubed_sphere#276
• Fix errors in fv_io.F90 NOAA-GFDL/GFDL_atmos_cubed_sphere#285

Issue(s) addressed

• fixes Unnecessary call to recover_fields for restart files can cause crashes. #694
• fixes CLM Lake restarts nine 3D fields that are constant ufs-community/ccpp-physics#112

Testing

How were these changes tested?

New or improved ufs-weather-model regression tests

What compilers / HPCs was it tested with?

jet, hera, cheyenne, hercules, orion, and cactus (wcoss2)

Are the changes covered by regression tests?

Some ufs-weather-model regression tests are added or modified.

Have the ufs-weather-model regression test been run? On what platform?

jet, hera, cheyenne, hercules, orion, and cactus (wcoss2)

Will the code updates change regression test baseline? If yes, why? Please show the baseline directory below.

Yes. All tests with FV3 will change due to NOAA-GFDL/GFDL_atmos_cubed_sphere#276

Please commit the regression test log files in your ufs-weather-model branch

Will do.

Dependencies

• waiting on Fixes to allow FV3_HRRR_c3 to run with gnu debug plus PR #113, #106, and #103 ufs-community/ccpp-physics#102
• waiting on initialize two arrays and fix fortran coding error plus PRs #285 and #276 NOAA-GFDL/GFDL_atmos_cubed_sphere#280

[SamuelTrahanNOAA]

I've merged #701 into this PR since the UFS code managers plan to merge both simultaneously.

Also, I've merged #697 into this PR since that one is expected to go in first.

[SamuelTrahanNOAA]

@haiqinli - Please confirm I merged your changes correctly.

[haiqinli]

@SamuelTrahanNOAA Your merging looks good to me. Thank you very much!

[grantfirl]

@SamuelTrahanNOAA ufs-community/ccpp-physics#102 was merged. Please prepare this one for merge.

[SamuelTrahanNOAA]

This PR is ready for merge.

• atmos_cubed_sphere is the head of the dev/emc branch of GFDL_atmos_cubed_sphere
• ccpp/physics is the head of the ufs/dev branch of ccpp-physics
• .gitmodules matches the current HEAD of develop