Should be of no consequence, right?

Add -fPIC to provided CFLAGS/FFLAGS if required

Fix thinko in skylake beta handling

casting ints is cheaper but it has a rounding, not memory casing effect, resulting in
invalid outcome

override user-provided FFLAGS if necessary

Add -mavx2 to TARGET=HASWELL builds

to leverage improvements from PR#1921

Reflect ARMV8 target definition changes from PR1876

and create config target directory for cross-compiles.

Avoid taking the root of a negative number in simple threaded syrk

Fixes #1924 where numpy 1.17+ would report the (transient) FE_INVALID exception raised for the domain error.

rebase

Replicate some of the SKYLAKEX dgemm improvements also to HASWELL

Haswell likes a GEMM_PREFERED_SIZE of 16 to improve the split that the
threading code does to make it a nice multiple of the SIMD kernel size

The dgemm_ncopy_8_skylakex.c code is not avx512 specific and gives
a nice performance boost for medium sized matrices

it's more efficient for certain tall/skinny matrices

(sgemm) Apply some of the SKYLAKEX optimizations also to HASWELL

sgemm_ncopy_4_skylakex.c uses SSE transpose operations where the
real perf win happens; this also works great for Haswell.

This gives double digit percentage gains on small and skinny matrices

with a few small changes it's possible to use the skylake sgemm code
also for haswell, this gives a modest gain (10% range) for smallish
matrixes but does wonders for very skinny matrixes

Add a "sgemm direct" mode for small matrixes

OpenBLAS has a fancy algorithm for copying the input data while laying
it out in a more CPU friendly memory layout.

This is great for large matrixes; the cost of the copy is easily
ammortized by the gains from the better memory layout.

But for small matrixes (on CPUs that can do efficient unaligned loads) this
copy can be a net loss.

This patch adds (for SKYLAKEX initially) a "sgemm direct" mode, that bypasses
the whole copy machinary for ALPHA=1/BETA=0/... standard arguments,
for small matrixes only.

What is small? For the non-threaded case this has been measured to be
in the M*N*K = 28 * 512 * 512 range, while in the threaded case it's
less, around M*N*K = 1 * 512 * 512

Fix for DYNAMIC_ARCH builds made on a AVX512-capable host

cf #1908

cf. #1908

Fix cmake parsing of GEMM kernels for ARMV8

The Makefile variable parser in utils.cmake currently does not handle conditionals. Having the definitions for non-OSX last will at least make cmake builds work again on non-OSX platforms.

Plain cgemm_kernel and zgemm_kernel are not used anywhere, only cgemm_kernel_b etc.
Needlessly building them (without any define like NN, CN, etc.) just happened to work on most platforms, but not on arm64. See #1870

must have happened in my #1449

Fix two mistakes on Arm64 builds