Lower byte swap intrinsic.

Clump the negate instruction w/ the bswap instruction as an
"inplace" operation. One difference is that bswap has stricter
requirements the operand type.

BUG= https://code.google.com/p/nativeclient/issues/detail?id=3882
R=stichnot@chromium.org, wala@chromium.org

Review URL: https://codereview.chromium.org/401533002