[X86_64] Use xor-with-self optimization for move(imm64, reg).
Created attachment 226324 [details] Patch