GNU bug fix for MOM_variables.F90

[mathomp4]

Upon the update of MOM6 in GEOS-ESM to the latest code, our CI GNU Debug testing threw a failure:

https://app.circleci.com/pipelines/github/GEOS-ESM/GEOSgcm/2849/workflows/cb295ace-751c-439c-9ad4-b1922a7c4cc7/jobs/5573/parallel-runs/0/steps/0-105

Tracing it back to line 1390 in src/parameterizations/vertical/MOM_diabatic_driver.F90

MOM6/src/parameterizations/vertical/MOM_diabatic_driver.F90

Lines 1389 to 1392 in b61b29e

	if (associated(visc%sfc_buoy_flx)) then
	visc%sfc_buoy_flx(:,:) = SkinBuoyFlux(:,:)
	call pass_var(visc%sfc_buoy_flx, G%domain, halo=1)
	endif

Experience with GNU tells us that GNU does not handle pointers that aren't nullified and not associated well. So associated(pointer) could be true, could be false, or unknown. I think the Fortran Standard says it's an undefined state.

The fix is simple though. Null the pointer in the derived type declaration. It also matches all the other pointers which are that way.

(For reference, this change came in via #1603 through NOAA-GFDL#352 it looks like.)

[codecov]

Codecov Report

Merging #1605 (f39e7f7) into main (b61b29e) will increase coverage by 0.00%.
The diff coverage is n/a.

❗ Current head f39e7f7 differs from pull request most recent head 82acb2f. Consider uploading reports for the commit 82acb2f to get more accurate results

@@           Coverage Diff           @@
##             main    #1605   +/-   ##
=======================================
  Coverage   38.16%   38.17%           
=======================================
  Files         269      269           
  Lines       76503    76503           
  Branches    14064    14064           
=======================================
+ Hits        29200    29203    +3     
+ Misses      42056    42052    -4     
- Partials     5247     5248    +1     

Files Changed	Coverage Δ
src/core/MOM_variables.F90	46.96% <ø> (ø)

... and 1 file with indirect coverage changes

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more

[kshedstrom]

I approve this PR.

[marshallward]

GFDL approves this change, and apologies for letting it slip through.

We will look over our testing procedure and try to apply any improvements (with -fcheck=all being the most apparent).

[jiandewang]

works fine in EMC

[marshallward]

Reminder to @alperaltuntas / @gustavo-marques, @abozec it review this, it should be relatively simple.

[abozec]

Sorry for the delay, I was out of the country. COAPS approves.

[marshallward]

Thanks all, and remember to update your forks!

[marshallward]

(Also this quasirandom fail should be fixed in the next dev/gfdl -> main update)

https://github.com/mom-ocean/MOM6/actions/runs/5833705621/job/15821701926