Update chgres_cube for fractional grids

[GeorgeGayno-NOAA]

DESCRIPTION OF CHANGES:

Update chgres_cube for fractional (partial land/non-land) grids. Update the format of the surface coldstart file (called v2). These changes are designed to be backwards compatible with the forecast model.

TESTS CONDUCTED:

If there are changes to the build or source code, the tests below must be conducted. Contact a repository manager if you need assistance.

• Compile branch on all Tier 1 machines using Intel (Orion, Jet, Hera and WCOSS2). Done using 45ca85a.
• Compile branch on Hera using GNU. Done using 45ca85a.
• Compile branch in 'Debug' mode on WCOSS2. Done using 45ca85a
• Run unit tests locally on any Tier 1 machine. Done on dogwood using 45ca85a. All passed.
• Run relevant consistency tests locally on all Tier 1 machine. Done using 45ca85a. All chgres_cube tests failed as expected. Most grid_gen tests failed as expected. For details, see notes below.

Describe any additional tests performed.

DEPENDENCIES:

NOAA-EMC/fv3atm#698
ufs-community/ufs-weather-model#2005

DOCUMENTATION:

N/A

ISSUE:

Fixes #123.

[BrianCurtis-NOAA]

@GeorgeGayno-NOAA If your FV3 PR NOAA-EMC/fv3atm#698 is ready to go, should this still be in draft form?

[GeorgeGayno-NOAA]

All grid_gen consistency tests failed except the c96 uniform test.

The sfc_climo_gen code was updated to always map surface fields to any point with at least some land (this is expected by updated chgres_cube). Previously, surface fields were mapped to points with at least some land only when the 'add_lake' option was used. Otherwise, surface fields were only mapped to points with 50% land.

The c96 uniform tests used the 'add_lake' option (and passed). The others did not use that option. As a result, the surface files from the remaining tests differed from the baseline (the 'grid' and 'oro' files were the same). As expected, the surface files created by the branch included additional data points at coasts. Inland, the data were the same.

[GeorgeGayno-NOAA]

A note about the chgres_cube consistency tests.

Since the updated chgres_cube requires valid surface values at all points with at least some land, the 'fixed' data input to the tests was updated.

The c96 and c192 tests now use the '202310127' version files ('grid', 'oro' and 'sfc' files).

The RRFS 13KM and 25KM fixed files ('grid', 'oro' and 'sfc' files) were recreated using the branch. Despite several attempts, I could not get the grids to exactly match (the CRES is slightly different). However, they are very close.

The RRFS 3KM surface fixed files were updated, but the same 'grid' and 'oro' files were used.

[GeorgeGayno-NOAA]

All chgres_cube consistency tests failed as expected.

Failures were due to:

• The new chgres outputs a different surface coldstart file format
• Differences in the input fixed data (new 'grid', 'oro' and 'sfc' files)
• The new chgres outputs data for a fractional land/non-land grid.

Files from the branch tests were visualized against the baseline data. The differences I saw were consistent with the updates for this PR and did not indicate a problem with the branch.