[msl] Polyfill signed integer arithmetic

Avoid UB caused by signed integer overflow by bitcasting to unsigned
integers and back again.

Bug: 42251016
Change-Id: I98078a9a4e9603b3ab6faa4dbc0fc0812a5dcc3e
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/205235
Reviewed-by: dan sinclair <dsinclair@chromium.org>
Commit-Queue: James Price <jrprice@google.com>

test/tint/array/assign_to_subexpr.wgsl.expected.ir.msl

diff --git a/test/tint/array/assign_to_subexpr.wgsl.expected.ir.msl b/test/tint/array/assign_to_subexpr.wgsl.expected.ir.msl
index 265bd44..c7b0190 100644
--- a/test/tint/array/assign_to_subexpr.wgsl.expected.ir.msl
+++ b/test/tint/array/assign_to_subexpr.wgsl.expected.ir.msl
@@ -34,7 +34,7 @@
   (*dst_ptr) = src;
   (*dst_struct_ptr).arr = src;
   (*dst_array_ptr)[0] = src;
-  return (((*dst_ptr)[0] + (*dst_struct_ptr).arr[0]) + (*dst_array_ptr)[0][0]);
+  return as_type<int>((as_type<uint>(as_type<int>((as_type<uint>((*dst_ptr)[0]) + as_type<uint>((*dst_struct_ptr).arr[0])))) + as_type<uint>((*dst_array_ptr)[0][0])));
 }
 
 kernel void tint_symbol(device int* s [[buffer(0)]]) {