Allow struct.wait expected value to be a supertype

[stevenfontanella]

For ref fields, the expected ref that's passed in does not need to be a subtype of the struct's field's type. See the typing rules from the proposal, and WebAssembly/shared-everything-threads#92 which explains why. From the spec:

wait_expected(i32) = i32
wait_expected(i64) = i64
wait_expected(t) = (ref null (shared eq)) -- if t <: (ref null (shared eq))

C |- struct.wait x y : [(ref null x) (ref null (shared waitqueue)) t i64] -> [i32]
 -- C.types[x] = shared? struct ft*
 -- ft*[y] = mut? t'
 -- t = wait_expected(t')

In this case t <: (ref null (shared eq)) and it's not required that t <: t'.

---

Since the expected value isn't required to be a subtype of the field type, we no longer need the fixing code in TypeRefining.cpp. This matches the behavior for the expected field of atomic cmpxchg, which doesn't touch curr->expected. The type-refining-gufa test was a mis-optimization before, because it caused the struct.wait to block (by changing the expected value to ref.null) when it did not block in the original program.

Part of #8315.