tint: Implement const eval of binary multiply

Bug: tint:1581
Change-Id: I70ff40ed4d8faf0a665824fef936ffbafb3f0948
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/99362
Kokoro: Kokoro <noreply+kokoro@google.com>
Commit-Queue: Antonio Maiorano <amaiorano@google.com>
Reviewed-by: Ben Clayton <bclayton@google.com>

test/tint/samples/function.wgsl.expected.dxc.hlsl

@@ -1,5 +1,5 @@
 float main() {
-  return (((2.0f * 3.0f) - 4.0f) / 5.0f);
+  return (2.0f / 5.0f);
 }
 
 [numthreads(2, 1, 1)]

test/tint/samples/function.wgsl.expected.fxc.hlsl

@@ -1,5 +1,5 @@
 float main() {
-  return (((2.0f * 3.0f) - 4.0f) / 5.0f);
+  return (2.0f / 5.0f);
 }
 
 [numthreads(2, 1, 1)]

test/tint/samples/function.wgsl.expected.msl

@@ -2,7 +2,7 @@
 
 using namespace metal;
 float tint_symbol() {
-  return (((2.0f * 3.0f) - 4.0f) / 5.0f);
+  return (2.0f / 5.0f);
 }
 
 kernel void ep() {

test/tint/samples/function.wgsl.expected.spvasm

@@ -1,7 +1,7 @@
 ; SPIR-V
 ; Version: 1.3
 ; Generator: Google Tint Compiler; 0
-; Bound: 16
+; Bound: 12
 ; Schema: 0
                OpCapability Shader
                OpMemoryModel Logical GLSL450
@@ -12,19 +12,15 @@
       %float = OpTypeFloat 32
           %1 = OpTypeFunction %float
     %float_2 = OpConstant %float 2
-    %float_3 = OpConstant %float 3
-    %float_4 = OpConstant %float 4
     %float_5 = OpConstant %float 5
        %void = OpTypeVoid
-         %12 = OpTypeFunction %void
+          %8 = OpTypeFunction %void
        %main = OpFunction %float None %1
           %4 = OpLabel
-          %7 = OpFMul %float %float_2 %float_3
-          %9 = OpFSub %float %7 %float_4
-         %11 = OpFDiv %float %9 %float_5
-               OpReturnValue %11
+          %7 = OpFDiv %float %float_2 %float_5
+               OpReturnValue %7
                OpFunctionEnd
-         %ep = OpFunction %void None %12
-         %15 = OpLabel
+         %ep = OpFunction %void None %8
+         %11 = OpLabel
                OpReturn
                OpFunctionEnd