AArch64 va_arg()

[Walter Bright]

Figuring out stdarg for the Arm.

```
#include <stdarg.h>
int square(float a, float b, ...)
{
     va_list args;
     va_start(args, b);
     int i = va_arg(args, int);
     va_end(args);
     return i;
}
```
which compiles to, I kid you not:
```
    0:	d10383ff 	sub	sp, sp, #0xe0   // function prolog
    4:	bd000fe0 	str	s0, [sp, #12]
    8:	bd000be1 	str	s1, [sp, #8]

    c:	f90053e0 	str	x0, [sp, #160]
   10:	f90057e1 	str	x1, [sp, #168]
   14:	f9005be2 	str	x2, [sp, #176]
   18:	f9005fe3 	str	x3, [sp, #184]
   1c:	f90063e4 	str	x4, [sp, #192]
   20:	f90067e5 	str	x5, [sp, #200]
   24:	f9006be6 	str	x6, [sp, #208]
   28:	f9006fe7 	str	x7, [sp, #216]
   2c:	3d8013e2 	str	q2, [sp, #64]
   30:	3d8017e3 	str	q3, [sp, #80]
   34:	3d801be4 	str	q4, [sp, #96]
   38:	3d801fe5 	str	q5, [sp, #112]
   3c:	3d8023e6 	str	q6, [sp, #128]
   40:	3d8027e7 	str	q7, [sp, #144] // initialize va_argsave

   44:	910383e0 	add	x0, sp, #0xe0
   48:	f9000fe0 	str	x0, [sp, #24]
   4c:	910383e0 	add	x0, sp, #0xe0
   50:	f90013e0 	str	x0, [sp, #32]
   54:	910283e0 	add	x0, sp, #0xa0
   58:	f90017e0 	str	x0, [sp, #40]
   5c:	128007e0 	mov	w0, #0xffffffc0            	// #-64
   60:	b90033e0 	str	w0, [sp, #48]
   64:	12800be0 	mov	w0, #0xffffffa0            	// #-96
   68:	b90037e0 	str	w0, [sp, #52]    // va_start(args, b); (filling in va_list)

   6c:	b94033e1 	ldr	w1, [sp, #48]
   70:	f9400fe0 	ldr	x0, [sp, #24]
   74:	7100003f 	cmp	w1, #0x0
   78:	540000ab 	b.lt	8c <square+0x8c>  // b.tstop
   7c:	91002c01 	add	x1, x0, #0xb
   80:	927df021 	and	x1, x1, #0xfffffffffffffff8
   84:	f9000fe1 	str	x1, [sp, #24]
   88:	1400000d 	b	bc <square+0xbc>
   8c:	11002022 	add	w2, w1, #0x8
   90:	b90033e2 	str	w2, [sp, #48]
   94:	b94033e2 	ldr	w2, [sp, #48]
   98:	7100005f 	cmp	w2, #0x0
   9c:	540000ad 	b.le	b0 <square+0xb0>
   a0:	91002c01 	add	x1, x0, #0xb
   a4:	927df021 	and	x1, x1, #0xfffffffffffffff8
   a8:	f9000fe1 	str	x1, [sp, #24]
   ac:	14000004 	b	bc <square+0xbc>
   b0:	f94013e2 	ldr	x2, [sp, #32]
   b4:	93407c20 	sxtw	x0, w1
   b8:	8b000040 	add	x0, x2, x0
   bc:	b9400000 	ldr	w0, [x0]
   c0:	b9003fe0 	str	w0, [sp, #60]   // int i = va_arg(args, int)

   c4:	b9403fe0 	ldr	w0, [sp, #60]   // return i

   c8:	910383ff 	add	sp, sp, #0xe0
   cc:	d65f03c0 	ret                     // epilog
```