Fix 1x1 and 2x2 blocksize error

[apcraig]

PR checklist

• Short (1 sentence) summary of your PR:
  Fix 1x1 and 2x2 blocksize error, update initialization of ARRAY_G in global_gather_ext

• Developer(s):
  apcraig

• Suggest PR reviewers from list in the column to the right.

• Please copy the PR test results link or provide a summary of testing completed below.
  Full test suite run on cheyenne with gnu. All results are bit-for-bit except 1x1 and 2x2 results which don't change answers but do change restart files slightly due to the fix in missing blocks. See cheyenne gnu results https://github.com/CICE-Consortium/Test-Results/wiki/cice_by_hash_forks#85531cf8f8ae93f628344df6ecfa4be12e9629d8

• How much do the PR code changes differ from the unmodified code?

  • bit for bit
  • different at roundoff level
  • more substantial

• Does this PR create or have dependencies on Icepack or any other models?

  • Yes
  • No

• Does this PR add any new test cases?

  • Yes
  • No

• Is the documentation being updated? ("Documentation" includes information on the wiki or in the .rst files from doc/source/, which are used to create the online technical docs at https://readthedocs.org/projects/cice-consortium-cice/. A test build of the technical docs will be performed as part of the PR testing.)

  • Yes
  • No, does the documentation need to be updated at a later time?
    • Yes
    • No

• Please provide any additional information or relevant details below:

• Fix 1x1 and 2x2 non-bit-for-bit results. It turns out the 1x1 and 2x2 blocks were
  bit-for-bit physically, just not technically. The problem was that a set
  of blocks were "land block eliminated"
  near the equator incorrectly. These blocks never have sea ice, so it's not a problem
  scientifically, but it changes the masking and restart files. The fix is a one line change
  in ice_domain.F90 which that sets the minimum "flat" value to 1. At the equator, flat
  was set to zero and this eliminated those gridcells incorrectly. This error could only
  happen in cases where the gridcells were within 0.5 degrees of the equator, and a set of
  conditions were required where an active gridcell was set to inactive which then resulted
  in the entire block being eliminated. This could really only happen with the smallest blocks
  which is why it only showed up with 1x1 and 2x2.

• Update the initialization of ARRAY_G in the gather_global_ext interfaces. Get rid of
  older, more complex approach which left some blocks uninitialized.

• Set psalt(n) to zero by default. Was uninitialized in cases where vice=0.

• Add a test to check bit-for-bit in log files for 1x1 blocks with bfbflag set to reprosum.

[apcraig]

I think this can close #595. This addresses the 1x1 and 2x2 block issue as well as the ARRAY_G issue. The evp1d solver is being addressed in #568.

It would be nice to merge this by Friday afternoon if we can so it can be included in weekend testing.

[eclare108213]

cool

[phil-blain]

I'm not that familiar with this part of the code, especially the MPI gather-scatter stuff, but removing a bunch of lines and replacing it with a single line is always welcome!

Thanks.