Choose type for integer literals to match the target mask size (if possible).

On a target with a 16-bit mask (for example), we would choose the type
of an integer literal "1024" to be an int16.  Previously, we used an int32,
which is a worse fit and leads to less efficient code than an int16
on a 16-bit mask target.  (However, we'd still give an integer literal
1000000 the type int32, even in a 16-bit target.)

Updated the tests to still pass with 8 and 16-bit targets, given this
change.

tests/test-85.ispc

@@ -8,7 +8,7 @@ float f(float a, int b) { return a + b; }
 float f(int i) { return i + 1.; }
 export void f_f(uniform float RET[], uniform float aFOO[]) {
     float a = aFOO[programIndex]; 
-    RET[programIndex] = f(a) + f() + f(a, a) + f(10); 
+    RET[programIndex] = f(a) + f() + f(a, a) + f(10l);
 }
 
 export void result(uniform float RET[]) {