Switch Mac GHA runner to latest available Intel system

[godotalgorithm]

Open PR to update the Mac GHA runner. I'll try to collapse the commits into something more reasonable once I get everything working, I've previously made messes in the Git logs in trying to debug GHA scripts...

Status

• Ready for merge

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 67.93%. Comparing base (8c64d1f) to head (12c96c7).
Report is 4 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #217   +/-   ##
=======================================
  Coverage   67.93%   67.93%           
=======================================
  Files         336      336           
  Lines       77025    77025           
=======================================
  Hits        52326    52326           
  Misses      24699    24699           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[godotalgorithm]

Eventually, the official MOPAC distribution on Mac will switch from Intel to ARM processors. Before that, I will probably switch all MOPAC distributions from Intel compilers to GNU or LLVM compilers. However, a major reason that I am holding off on this change for as long as possible is that the MKL symmetric eigensolver has an appreciable performance advantage (nearly 2x) over the reference LAPACK implementation used by OpenBLAS. The symmetric eigensolver is one of the main bottlenecks of MOPAC, so moving away from MKL will have an immediate performance impact if future changes to reference LAPACK, OpenBLAS, or some other vendor-specific version of LAPACK do not replicate the (still-proprietary) performance gains of MKL's symmetric eigensolver. Unfortunately, I'm not optimistic about this, and MOPAC will probably suffer a permanent performance drop when it is eventually forced away from using MKL.