Update components.yaml to match GEOSgcm v10.22.1 (and update CI)

[mathomp4]

This PR updates ESMA_env, ESMA_cmake, and MAPL in the GEOSldas to match GEOSgcm v10.22.1. The changes are:

• ESMA_env
  • Update to Baselibs 6.2.13 (v3.12.0)
  • Move to have both Python2 and Python3 loaded at the same time (v3.12.0)
  • Minor tweak to parallel_build.csh (v3.13.0)
• ESMA_cmake
  • Updates for Spack support (v3.11.0)
  • Preliminary M1 support (v3.12.0)
• MAPL
  • well...A lot of bug fixes (see https://github.com/GEOS-ESM/MAPL/blob/main/CHANGELOG.md)

Of these, the ESMA_cmake is needed for build issues as GMAO_Shared main needs this version.

Note that there is a chance the MAPL update could be sort-of non-zero-diff. The issue is that Ops found some lats and lons in history output were of the sort of "6.500000000001" instead of "6.5". This MAPL update "fixes" that, and comparators like nccmp will see this as a difference. The data is zero-diff, but the lats and lons are not.

[biljanaorescanin]

I will test it now, thanks

[biljanaorescanin]

Tests globalcs, aggglobalcs and gnuglobalcs fail for model and assim run comparison, for both collections ("lnd" and "lfs" ).
If we compare these tests without lat/lon files are identical. ( nccmp -dfsB -x lat,lon )

[gmao-rreichle]

Tests globalcs, aggglobalcs and gnuglobalcs fail for model and assim run comparison, for both collections ("lnd" and "lfs" ). If we compare these tests without lat/lon files are identical. ( nccmp -dfsB -x lat,lon )

How about the other tests (conus, global, cnclm[xx])? I assume you're implying that the other tests passed?

[biljanaorescanin]

Yes, all other tests passed.

Runtype Clone Build Build Time Model Run/Compare Assim Run/Compare

---

conus pass pass 14 min pass/pass -- / --
global -- -- -- pass/pass pass/pass
globalcs -- -- -- pass/FAIL pass/FAIL
globalcnclm4 -- -- -- pass/pass -- / --
globalcnclm45 -- -- -- pass/pass -- / --
debugconus -- pass 5 min pass/pass -- / --
aggconus -- pass 15 min pass/pass -- / --
aggglobal -- -- -- pass/pass pass/pass
aggglobalcs -- -- -- pass/FAIL pass/FAIL
aggglobalcnclm4 -- -- -- pass/pass -- / --
aggglobalcnclm45 -- -- -- pass/pass -- / --
gnuconus pass pass 12 min pass/pass -- / --
gnuglobal -- -- -- pass/pass pass/pass
gnuglobalcs -- -- -- pass/FAIL pass/FAIL
gnuglobalcnclm4 -- -- -- pass/pass -- / --
gnuglobalcnclm45 -- -- -- pass/pass -- / --
gnudebugconus -- pass 10 min pass/pass -- / --

[mathomp4]

I know it's merged, but I think that makes sense that only the CS tests showed a difference.