GEOSgcm GNU Debug no longer runs after MAPL 2.48

[mathomp4]

Runs of GEOSgcm with GNU (13 or 14) are now failing with:

        EXTDATA: DEBUG: ExtData Run_: READ_LOOP: Done
[borgj101:222883] *** An error occurred in MPI_Wait
[borgj101:222883] *** reported by process [2176581633,0]
[borgj101:222883] *** on communicator MPI COMMUNICATOR 22 CREATE FROM 21
[borgj101:222883] *** MPI_ERR_TRUNCATE: message truncated
[borgj101:222883] *** MPI_ERRORS_ARE_FATAL (processes in this communicator will now abort,
[borgj101:222883] ***    and potentially your MPI job)
>> Error << /discover/swdev/gmao_SIteam/MPI/openmpi/4.1.6-SLES15/gcc-13.2.0/bin/mpirun  -np 96 /discover/nobackup/mathomp4/Experiments/stock-2024Oct04-1day-c24-GNU-DEBUG-ExtDataDebug/scratch/GEOSgcm.x --logging_config logging.yaml: status = 15; at /gpfsm/dnb34/mathomp4/SystemTests/builds/AGCM_GNU/CURRENT/GEOSgcm/install-Debug/bin/esma_mpirun line 377.
GEOSgcm Run Status: -1

So after ExtData runs, we crash. Running with DDT showed it crashing here:

MAPL/base/MAPL_EsmfRegridder.F90

Lines 1344 to 1351 in d543cf8

	if (spec%regrid_method /= REGRID_METHOD_NEAREST_STOD) then
	call ESMF_FieldRegrid(src_field, dst_field, &
	& routeHandle=route_handle, &
	& dynamicMask=this%dynamic_mask, &
	& termorderflag=ESMF_TERMORDER_SRCSEQ, &
	& zeroregion=ESMF_REGION_SELECT, &
	& rc=status)
	_VERIFY(status)

with an error code of ESMC_RC_NOT_IMPL (which seems to be the catchall of ESMF_FieldRegrid failure).

Now, tracking down in my nightly tests when GNU develop Debug runs failed I happed on the same exact time @bena-nasa converted ExtData to containers and things became non-zero-diff (see #3025), aka PR #3007

So, as a final test, I did a cherry-pick of commit d888902 (aka before #3007):

git checkout d888902b764e3c5d6cf7971eec2fe766ff3b1d2a -- gridcomps/ExtData2G/CMakeLists.txt
git checkout d888902b764e3c5d6cf7971eec2fe766ff3b1d2a -- gridcomps/ExtData2G/ExtDataGridCompNG.F90
git checkout d888902b764e3c5d6cf7971eec2fe766ff3b1d2a -- gridcomps/ExtData2G/ExtDataOldTypesCreator.F90

and that does run. So something between d888902...cae216f GNU does not like. :(

[mathomp4]

Note @tclune reminded me of ESMF Logging and I did try that but all we saw were the errors due to #1976 (which we should fix as well).