proposal: compiler specific vectorization ( e.g. `__builtin_shufflevector` ... ) if native SIMD is not ( could not ) detected

[recp]

We can use builtin compiler vectorization ( e.g. __builtin_shufflevector ... ) helpers ( if available ) as fallback of SIMD to native ( if possible ):

https://gcc.gnu.org/onlinedocs/gcc/Vector-Extensions.html https://clang.llvm.org/docs/LanguageExtensions.html#langext-builtin-shufflevector