Update gdma to v2.3; convert to runtime

[loriab]

Description

User API & Changelog headlines

• gdma is bumped from upstream v2.2.06 to v2.3.3 (can now use h-functions)
• gdma is now selectable at runtime, not just compile-time. use conda install pygdma -c conda-forge (presently in staged-recipes adding gdma conda-forge/staged-recipes#22882 )

Dev notes & details

• A perhaps obscure point is that psi hasn't been using exactly upstream GDMA, which is a Fortran executable. Instead, Andy Simmonett did lots of I/O conversion and turned it into a library and bound some functions to C so psi could extract results w/o parsing. That's great, but it's meant we've been fixed at upstream v2.2.06.
• I was startled to realize gdma is GPL. For background, roughly speaking, GPL probably does adhere to psi4 if libgdma statically linked, might adhere if dynamically linked (FSF thinks yes; everyone else thinks no; result is that any GPL makes ppl nervous), doesn't adhere if dlopen'ed or runtime optional. Conda-package-wise, we've always dynamically linked to libgdma, and scouring my memory, there was a plan to have an alternate gdma- and chemps2-free conda build of psi4 to remove doubt; that never happened.
• To get past these problems and get a c-f package going, I
  • Started with current upstream https://gitlab.com/anthonyjs/gdma/-/tree/master and applied Andy's and Holger's gdma->libgdma changes
  • Updated the cmake buildsys with the usual checklist cmake 2023 updates gdma#6
  • Added back the gdma executable so c-f package wouldn't lose functionality from upstream
  • pybind11-'d the C functions Andy created
  • psi4-side, moved the gdma_interface c-side function over to the driver, so it's pure runtime
  • got the conda package going adding gdma conda-forge/staged-recipes#22882 for L, M, W
• need to add comment notation so the driver qcvars are associated with gdma
• UNRELATED: fix mrcc addon so that can pass it FNO orbitals again.

Checklist

• Tests added for any new features
• All or relevant fraction of full tests run

Status

• Ready for review
• Ready for merge

[JonathonMisiewicz]

Do we have a test on the distributed multipoles?

[loriab]

Do we have a test on the distributed multipoles?

Just one: https://github.com/psi4/psi4/blob/master/tests/gdma/gdma1/input.dat that's mirrored at https://github.com/psi4/psi4/blob/master/tests/pytests/test_addons.py#L14