AstGen: fix loop result locations

The main problem was that the loop body was treated as an expression
that was one of the peer result values of a loop, when in reality the
loop body is noreturn and only the `break` operands are the result
values of loops.

This was solved by introducing an override that prevents rvalue() from
emitting a store to result location instruction for loop bodies.

An orthogonal change also included in this commit is switching
`elem_val` index expressions to using `coerced_ty` and doing the
coercion to `usize` inside `Sema`, resulting in smaller ZIR (since the
cast becomes implied).

I also changed the break operand expression to use `reachableExpr`,
introducing a new compile error for double break.

This makes a few more behavior tests pass for `while` and `for` loops.

src/Sema.zig

@@ -12528,7 +12528,7 @@ fn elemVal(
     block: *Block,
     src: LazySrcLoc,
     array: Air.Inst.Ref,
-    elem_index: Air.Inst.Ref,
+    elem_index_uncasted: Air.Inst.Ref,
     elem_index_src: LazySrcLoc,
 ) CompileError!Air.Inst.Ref {
     const array_src = src; // TODO better source location
@@ -12538,6 +12538,8 @@ fn elemVal(
         return sema.fail(block, src, "array access of non-indexable type '{}'", .{array_ty});
     }
 
+    const elem_index = try sema.coerce(block, Type.usize, elem_index_uncasted, elem_index_src);
+
     switch (array_ty.zigTypeTag()) {
         .Pointer => switch (array_ty.ptrSize()) {
             .Slice => {