POWER10:  Optimize dgemv_n

Handling as 4x8 with vector pairs gives better performance than
existing code in POWER10.

Restore RISCV entries accidentally trashed by my PR 3005

add Fujitsu compilers

Co-authored-by: Tomoki Karatsu <karatsu.spack@gmail.com>

Add -msse for x86 and silence build warning in getarch

ARM64 DYNAMIC_ARCH build fix for BSD/OSX

Ensure that all targets in a DYNAMIC_ARCH build on POWER use the same buffer size

Fix ambiguous ifdefs in tests for user-defined options in Makefiles

rebase

allow setting soname without suffix or prefix

Improve the performance of dasum and sasum when SMP is defined

Add reproducer test for crash after fork

Don't overwrite blas_thread_buffer if already set

See #2993 for an analysis

After a fork it is possible that blas_thread_buffer has already
allocated memory buffers: goto_set_num_threads does allocate those
already and it may be called by num_cpu_avail in case the OpenBLAS
NUM_THREADS differ from the OMP num threads.
This leads to a memory leak which can cause subsequent execution of BLAS
kernels to fail.

Fixes #2993

If the toolchain is too old, or the mma features isn't set on a POWER10
fall back to the POWER9 loops.

LDVERSIONGTEQ35 needs to escape the '>' character.

LDVERSIONGTEQ35 is checking the system ld version which may be different
to the toolchain being used to compile OpenBLAS. We don't have a path
to the linker in our Makefiles, so (ab)use gcc -Wl,--version to get the
version of ld in our toolchain.

Fix sum optimize issues

Optimize the performance of rot by using universal intrinsics

Fix missing -mfma compiler flag in cmake builds without DYNAMIC_ARCH

rebase