Use GCC builtins for bswap as compiler can then emit these via insn

Using current macros, one gets for:

static inline int BSWAP32(int x) {
	return ((x << 24) | ((x << 8) & 0x00FF0000) | ((x >> 8) & 0x0000FF00) | (x >> 24));
}
int main(int argc, char **argv)
{
  return BSWAP32(argc);
}

main:
.LFB1:
	.cfi_startproc
	movl	%edi, %eax
	movl	%edi, %edx
	sarl	$24, %edx
	sall	$24, %eax
	orl	%edx, %eax
	movl	%edi, %edx
	sarl	$8, %edi
	sall	$8, %edx
	andl	$65280, %edi
	andl	$16711680, %edx
	orl	%edx, %eax
	orl	%edi, %eax
	ret

while using:

```c
int main(int argc, char **argv)
{
  return __builtin_bswap32(argc);
}

one gets:

main:
.LFB0:
	.cfi_startproc
	movl	%edi, %eax
	bswap	%eax
	ret