* Added ARMV9SME target
* Added SGEMM_DIRECT kernel based on SME1

Remove any optimization flags from DEBUG builds on POWER architecture

Add dummy implementations of openblas_get/set_affinity for OpenMP builds

Update Cray compiler options and calling convention in CMake

Corrections for Cray and Nvidia Fortran compiler calling conventions

Add initial autodetection support for Apple M4 as VORTEX with HAVE_SME

Fix misleading indentation warning in getrf

CI: Update upload-artifacts version in the homebrew workflow again

explicitly link to OpenMP

Provide copies of GEMMT under its new name GEMMTR adopted by the Reference BLAS

x86_64: opt somatcopy_ct with AVX

gh-actions: Test LoongArch64 with gcc14 from Ubuntu 24.04

Remove some unused inline macro definitions

fix: return types of *trtrs routines

cmake: set `CMP0042` to `NEW`

Optimize gemv forwarding on ARM64 systems

Fix AIX build

work around mingw32-gfortran 14.2 miscompiling CBLAS1 tests

Properly report HAVE_SVE in ARM64 autodetection where applicable

LoongArch64: Opt somatcopy with LASX

Update README.md

[POWER] Vectorize BF16 GEMV