Update to HDF5 1.14.4

[mathomp4]

I think it's finally time we move to HDF5 1.14.3. This might cause some agita with the land group (cc @jardizzo and @gmao-rreichle) but the fact that the HDF Group doesn't even list HDF5 1.10 on their snazzy new download page anymore, makes me think we've reached the end.

It will probably be its own minor release of ESMA-Baselibs itself, but it needs to be done.

That said, I know SMAP or one of the land outputs might still be in some ancient version of HDF5 (1.8?) so this might take a few tries.

[gmao-rreichle]

@mathomp4 - sure, why not try. Should we fold it into the ongoing update of repo versions in GEOS-ESM/GEOSldas#681? If you have a new Baselibs version, we can just try the GEOSldas nightly tests, which include reading of SMAP h5 files. If that works, we should be all set for L4_SM. @jardizzo should weigh in on testing the SMAP L4 Carbon algorithm, which is not in GEOS-ESM.
cc: @biljanaorescanin

[mathomp4]

@gmao-rreichle I don't have a version of Baselibs with it yet. ESMF update was the major one. I'll look at getting one soon, maybe next week. I know I tested HDF5 1.12 a while back and GEOSgcm didn't care, so I'm hoping 1.14 is fine as well.

[mathomp4]

Reopening due to issues being seen with nf90_create.