Modrenize R benchmarks slightly

Revert generate to install, explain the nature of the affinity conflict

Lots of small changes in the wording of the comments, plus an expansion of the NUM_THREADS and NO_AFFINITY sections.

Change ARMV8 target with BINARY=32 to ARMV7 automatically

fixes #1961

fixes #1984

Shift transition to multithreading towards larger matrix sizes

Report SkylakeX as Haswell if compiler does not support AVX512

... or make was invoked with NO_AVX512=1

Increase Zen SWITCH_RATIO to 16

fixes #1974

following GEMM benchmarks on Ryzen2700X. For #1464

See #1886 and JuliaRobotics issue 500. trsm benchmarks on Haswell and Zen showed that with these values performance is roughly doubled for matrix sizes between 8x8 and 14x14, and still 10 to 20 percent better near the new cutoff at 32x32.

* Tag arguments 0 and 1 as both input and output

* Tag arguments 0 and 1 as both input and output (see #1964)

* Tag operands 0 and 1 as both input and output

For #1964 (basically a continuation of coding problems first seen in #1292)

crot fix

Blas1 single missing kernels implemented with vector builtins

Fixed idamin,icamin choosing the first occurance index of equal minimals

Add support for Hygon Dhyana

Move TLS key deletion to openblas_quit