Update from develop for 0.3.21 release

Update Changelog for 0.3.21

Increase thresholds for STFSM and CTFSM in the LAPACK testsuite

Update LAPACK/LAPACKE to Reference-LAPACK 3.10.1

Add more x86_64 target definitions for CMAKE cross-compiling

* Add support for the Cray Fortran compiler

Add TCORE Generic

Add env variable OMP_ADAPTIVE to control OMP threadpool behaviour

corrected indentation of for and if statement dgemv_thread_safety.cpp

During the cross-compilation on x64 host with MSVC for ARMv8, the
build fails as there is no define directives for Generic core.

Fix crash in RISCV autodetection when pmodel is not present in /proc/cpuinfo

Acknowledge past CZI EOSS 1/EOSS 3 funding

LoongArch64: Add DYNAMIC_ARCH support

POWER: Enable bfloat16 kernels by default

This patch enables bfloat16 kernels by default for POWER processors.
Tested on Linux POWER8, POWER9, POWER10 and AIX POWER10 systems.

Add openblas_getaffinity() extension (Linux-only)

* fix detection of Neoverse V1 and user-enforced selection of N2

Neoverse N2: DYNAMIC_ARCH

Add c_check recognition of Fujitsu fcc for Fugaku A64FX

LoongArch64: Add core LOONGSON2K1000 and LOONGSONGENERIC

utest needs to be linked against libm on QNX as well

SPARC: fix DNRM2 returning INF instead of zero due to intermediate overflow

PPC6: Fix DNRM2 returning INF instead of zero due to intermediate overflow

Add back travis_wait to keep ppc jobs from getting cancelled

POWER: Fix complex dot function failures

POWER10: dgemv builtin rename