Add separate intrinsics for MMX / SSE shifts with i32 integer operands. This allow...