Update utilities versions (components.yaml)

[gmao-rreichle]

This PR updates GEOSldas to the latest versions of the utilities in preparation for next release:

ESMA_cmake  v3.41.0
GMAO_Shared v1.9.7
GEOS_Util   v2.0.7
MAPL        v2.44.0

Note: ESMA_env v4.26.0 does not work, see #726

0-diff tests by @biljanaorescanin all passed.

Please let me know if you have any concerns re. the above versions.

@biljanaorescanin @weiyuan-jiang @mathomp4

[mathomp4]

I think those are all good to me.

Note that eventually I can work with @weiyuan-jiang to add the (new style) MPI detection into lenkf.j so that you can set MPI flags/variables depending on the stack (like how the GCM does for Intel MPI, Open MPI, MPT...)

[gmao-rreichle]

I think those are all good to me

Thanks, @mathomp4. Alas, I'm getting a floating point exception early in the LDAS_DEBUGCONUS/model run. The GEOSldas log file (./scratch/GEOSldas_log_txt) suggests it may be in the forcing initialization, but unfortunately it's not clear immediately where the crash occurs. Interestingly, GNUDEBUGCONUS runs fine.
I'm not sure what in the new utilities may be causing this. I can try to roll back the utilities updates one by one when I'm back in the office tomorrow.
After taking another look at my comment in #713, I realize I might have been to optimistic and may not have run the debug tests because those are model-only simulations and don't use HDF5. My thinking was probably that we're ok as long as the standard optimization works for Intel and GNU with the "assim" test case that exercises HDF5. But there may be some other change in ESMA_env that broke the debug simulation -- which is likely poor coding in GEOSldas revealed by debug. Oh my.

[biljanaorescanin]

All tests passed.