Enable -mbmi2

[tzaeschke]

Add the -mbmi2 for compiling with clang/gcc with cmake/bazel.
BMI2 enables use of CLZ (count leading zeos) which is used by the array_map.
Also: check for corresponding option in MSVC.

Alternatively:
build:linux --copt="-march=haswell" / skylake
build:linux --copt="-march=native"

https://learn.microsoft.com/en-us/cpp/intrinsics/bitscanreverse-bitscanreverse64?view=msvc-170