(Re)apply Makefile fix for supporting only a subset of precision types

Correct constant ALPHA in ?ORBDB6 (Reference-LAPACK PR 928)

Fix accumulation in LAPACK ?LASSQ (Reference-LAPACK PR 909)

Fix potential integer overflow in LAPACK C/ZBDSQR (Reference-LAPACK PR 927)

* Clean up ZROTG

Use SkylakeX ?ASUM microkernel for Cooperlake/Sapphirerapids as well

Fix outdated KERNEL files for SVE Cortex by aliasing to ARMV8SVE

Add dynamic dispatch to AIX and clang for Power

CirrusCI: Add FreeBSD clang/gfortran build with OpenMP

Increase the default GEMM buffer size on modern ARM server cpus

Clarify make/make install in the README and update the TARGET list there

POWER: AIX: Make use of power10 optimization

POWER10 optimizations are disabled when using default AIX assembler.
As we have fixed many issues recently, enabling optimization path
for default assembler.

Run nightly Homebrew cron job only on the main repo, not on forks

I noticed this because GitHub emailed me that it would disable the
nightly job because it hadn't changed for 3 months. It currently takes
30-50 minutes daily, and by default runs on all forks of the main
repository that have the relevant workflow yaml file. That serves little
purpose and wastes quite a bit of energy - so disable the runs outside
of the main repo.

This will not disable the runs on forks already made in the past that
contain this workflow file, but it does save 3 months worth of runs on
every new fork that is created.

[skip ci]

GH Workflows: Switch MINGW-W64 jobs to UCRT

Fix unwanted "hard" fallback to Prescott in runtime detection of Intel cpus

Fix compilation with (the fortran compiler from) Cray CCE

POWER: Increase macro size limit for AIX

Fix bfloat16_bits union so that it always the sizeof unsigned short for AIX.

AzureCI: move OSX-Clang jobs to macOS-12 to resolve setup/build timeouts

This patch increases the macro size limit from 4096 to 16384 to
allow compiling larger assembly files in AIX.
Tested with GCC and IBM Open XL C.

POWER10: Fix compilation issues with Open XL C

Fix build on Fujitsu A64FX