Update to HDF5 1.14.3, curl 8.6.0, zlib 1.3.1

[mathomp4]

Closes #137
Closes #155
Closes #154

Updates ESMA-Baselibs to use HDF5 1.14.3

[mathomp4]

All tests show this is zero-diff with GEOSgcm. We just need to bite the bullet. HDF5 1.10 had its day.

Plus I imagine the the day after I move to 1.14.3, @byrnHDF will release 1.14.4 or 1.16 or something. This is the way releases are born in the software world. 😆

[mathomp4]

Oh. Need to test on macOS as well. Never assume!

[byrnHDF]

See https://github.com/HDFGroup/hdf5/blob/develop/README.md, 1.14.4 is scheduled for release at end of March.

Allen

[mathomp4]

See https://github.com/HDFGroup/hdf5/blob/develop/README.md, 1.14.4 is scheduled for release at end of March.

Allen

Welp, no excuses for me to be lazy then! Thanks!

[mathomp4]

@byrnHDF Just a notice in case you ever happen back on this repo, we are reverting to HDF5 1.10 for a bit. We are finding weird issues where we cannot run nf90_create with various compilers with various settings.

For example, Intel ifort 2021.10 + some debug flags has been causing odd failures in our testing. I can even cause the issue with a simple netCDF reproducer.

And today, I encountered the same issue with NAG Fortran and nf90_create.

We are using the latest netcdf C and Fortran which say they have HDF5 1.14 support. And, weirdly, Spack with HDF5 1.14 seems to work just fine. Our hdf5 build flags aren't very exciting:

ESMA-Baselibs/GNUmakefile

Lines 540 to 552 in 90e2c74

	hdf5.config :: hdf5/README.md szlib.install zlib.install
	echo Configuring hdf5
	(cd hdf5; \
	export PATH="$(prefix)/bin:$(PATH)" ;\
	export LIBS="-lm" ;\
	./configure --prefix=$(prefix) \
	--includedir=$(prefix)/include/hdf5 \
	--with-szlib=$(prefix)/include/szlib,$(prefix)/lib \
	--with-zlib=$(prefix)/include/zlib,$(prefix)/lib \
	--disable-shared --disable-cxx\
	--enable-hl --enable-fortran --disable-sharedlib-rpath \
	$(ENABLE_GPFS) $(H5_PARALLEL) $(HDF5_ENABLE_F2003) \
	CFLAGS="$(CFLAGS) $(HDF5_NCCS_MPT_CFLAG)" FCFLAGS="$(NAG_FCFLAGS)" CC=$(NC_CC) FC=$(NC_FC) CXX=$(NC_CXX) F77=$(NC_F77) )

so... 🤷🏼

For now reversion is our safest path until we can figure this out (or when we move to spack and no longer seem to care for some reason)