Add reduce_add() for int8 and int16 types.

This maps to specialized instructions (e.g. PSADBW) when available.

tests/reduce-add-int16.ispc

@@ -0,0 +1,21 @@
+
+export uniform int width() { return programCount; }
+
+
+
+export void f_fu(uniform float RET[], uniform float aFOO[], uniform float b) {
+    float v = aFOO[programIndex];
+    uniform float m;
+    int16 iv = (int)v;
+/*CO    if (iv & 1)*/
+        m = reduce_add(iv);
+    RET[programIndex] = m;
+}
+
+export void result(uniform float RET[]) { 
+    uniform int x = 0;
+    for (uniform int i = 1; i <= programCount; ++i)
+        x += i;
+    RET[programIndex] = x;
+}
+