Array of array: Implement the core functionality: types, constructors, operations.

There will be subsequent commits to refine semantics, esp. version-specific semantics,
as well as I/O functionality and restrictions.

Note: I'm getting white-space differences in the preprocessor test results,
which I'm not checking in.  I think they need to be tagged as binary or something.

Test/430AofA.frag

@@ -0,0 +1,108 @@
+#version 430
+
+float[4][5][6] many[1][2][3];
+
+float gu[][7];
+float gimp[][];    // ERROR, implicit inner
+float g4[4][7];
+float g5[5][7];
+
+float[4][7] foo(float a[5][7])
+{
+    float r[7];
+    r = a[2];
+    float[](a[0], a[1], r, a[3]);              // ERROR, too few dims
+    float[4][7][4](a[0], a[1], r, a[3]);       // ERROR, too many dims
+    return float[4][7](a[0], a[1], r, a[3]);
+    return float[][](a[0], a[1], r, a[3]);
+    return float[][7](a[0], a[1], a[2], a[3]);
+}
+
+void bar(float[5][7]) {}
+
+void main()
+{
+    {
+        float gu[3][4][2];
+
+        gu[2][4][1] = 4.0;                     // ERROR, overflow
+    }
+    vec4 ca4[3][2] = vec4[][](vec4[2](vec4(0.0), vec4(1.0)),
+                              vec4[2](vec4(0.0), vec4(1.0)),
+                              vec4[2](vec4(0.0), vec4(1.0)));
+    vec4 caim[][2] = vec4[][](vec4[2](vec4(4.0), vec4(2.0)),
+                              vec4[2](vec4(4.0), vec4(2.0)),
+                              vec4[2](vec4(4.0), vec4(2.0)));
+    vec4 caim2[][] = vec4[][](vec4[2](vec4(4.0), vec4(2.0)),
+                              vec4[2](vec4(4.0), vec4(2.0)),
+                              vec4[2](vec4(4.0), vec4(2.0)));
+    vec4 caim3[3][] = vec4[][](vec4[2](vec4(4.0), vec4(2.0)),
+                               vec4[2](vec4(4.0), vec4(2.0)),
+                               vec4[2](vec4(4.0), vec4(2.0)));
+
+    vec4 a4[3][2] = {vec4[](vec4(0.0), vec4(1.0)),
+                     vec4[2](vec4(0.0), vec4(1.0)),
+                     vec4[2](vec4(0.0), vec4(1.0)) };
+    vec4 aim[][2] = {vec4[2](vec4(4.0), vec4(2.0)),
+                     vec4[](vec4(4.0), vec4(2.0)),
+                     vec4[2](vec4(4.0), vec4(2.0)) };
+    vec4 aim2[][] = {vec4[2](vec4(4.0), vec4(2.0)),
+                     vec4[2](vec4(4.0), vec4(2.0)),
+                     vec4[](vec4(4.0), vec4(2.0)) };
+    vec4 aim3[3][] = {vec4[2](vec4(4.0), vec4(2.0)),
+                      vec4[2](vec4(4.0), vec4(2.0)),
+                      vec4[2](vec4(4.0), vec4(2.0)) };
+
+    vec4 bad2[3][] = {vec4[2](vec4(4.0), vec4(2.0)),              // ERROR
+                      vec4[3](vec4(4.0), vec4(2.0), vec4(5.0)),
+                      vec4[2](vec4(4.0), vec4(2.0)) };
+
+    vec4 bad3[3][] = {vec4[3](vec4(4.0), vec4(2.0), vec4(5.0)),   // ERROR
+                      vec4[2](vec4(4.0), vec4(2.0)),
+                      vec4[2](vec4(4.0), vec4(2.0)) };
+
+    vec4 bad4[4][] = {vec4[2](vec4(4.0), vec4(2.0)),              // ERROR
+                      vec4[2](vec4(4.0), vec4(2.0)),
+                      vec4[2](vec4(4.0), vec4(2.0)) };
+
+
+    g4 = foo(g5);
+    g5 = g4;           // ERROR, wrong types
+    gu = g4;           // ERROR, not yet sized
+
+    foo(gu);           // ERROR, not yet sized
+    bar(g5);
+
+    if (foo(g5) == g4)
+        ;
+    if (foo(g5) == g5)  // ERROR, different types
+        ;
+
+    float u[][7];
+    u[2][2] = 3.0;
+    float u[5][7];
+    u[5][2] = 5.0;      // ERROR
+    foo(u);
+}
+
+void foo3()
+{
+    float resize1[][5][7];
+    resize1.length();           // ERROR
+    resize1[1][4][5] = 2.0;
+    resize1.length();           // ERROR
+    float resize1[3][5][7];
+    resize1.length();           // 3 in AST
+    resize1[1].length();        // 5 in AST
+    resize1[1][1].length();     // 7 in AST
+    resize1[1][1][1].length();  // ERROR
+
+    float resize2[][5][7];
+    float resize2[3][4][7];     // ERROR, inner dim change
+
+    float resize3[][5][7];
+    float resize3[3][5][9];     // ERROR, inner dim changed
+
+    float resize4[][5][7];
+    int  resize4[3][5][7];      // ERROR, element type
+}