Optimize _byteCount with compiler intrinsics

[AZero13]

Use compiler intrinsics to make _byteCount faster

[AZero13]

@bnbarham @DougGregor Thoughts on this?

[AZero13]

@FranzBusch Thoughts? Could you please run please test

[compnerd]

There is no guarantee that this is faster. The loop idiom recognizers will often make the appropriate optimizations. This simply makes the code less portable - __builtin_ctz is not a valid directive for MSVC.

[AZero13]

There is no guarantee that this is faster. The loop idiom recognizers will often make the appropriate optimizations. This simply makes the code less portable - __builtin_ctz is not a valid directive for MSVC.

I can show you on compiler explorer.