Add routines to standard library to do efficient AOS/SOA conversions.

Currently, we just support 3 and 4-wide variants (i.e. xyzxyz.. and xyzwxyzw..),
for int32 and float types.

tests/aossoa-2.ispc

@@ -0,0 +1,28 @@
+
+export uniform int width() { return programCount; }
+
+export void f_v(uniform float RET[]) {
+#define width 4
+#define maxProgramCount 16
+//CO    const uniform int width = 4;
+//CO    const uniform int maxProgramCount = 16;
+    uniform float a[width*maxProgramCount], r[width*maxProgramCount];
+    for (uniform int i = 0; i < width*maxProgramCount; ++i)
+        a[i] = i;
+
+    float x=-1, y=-1, z=-1, w=-1;
+    aos_to_soa4(a, 0, x, y, z, w);
+
+    int errs = 0;
+    if (x != width * programIndex) ++errs;
+    if (y != 1 + width * programIndex) ++errs;
+    if (z != 2 + width * programIndex) ++errs;
+    if (w != 3 + width * programIndex) ++errs;
+
+    RET[programIndex] = errs; 
+}
+
+export void result(uniform float RET[]) {
+    RET[programIndex] = 0;
+}
+