wfcheck checks where-clauses post normalization

[lcnr]

https://github.com/rust-lang/rust/blob/90b65889799733f21ebdf59d96411aa531c5900a/compiler/rustc_hir_analysis/src/check/wfcheck.rs#L1502-L1505

This normalize call is a noop in the new solver. I think we should just slap a deeply_normalize there instead and add a FIXME that we should check where-clauses for WF pre normalization. Though separately, we should do a crater run to see the impact of not normalizing here. The below breakage simply seems desirable to me and idk if there's non desirable breakage

---

rows 609 and 610, modcholesky crater logs, godbolt

pub struct View<A>(A);
pub trait Data {
    type Elem;
}
impl<'a, A> Data for View<&'a A> {
    type Elem = A;
}

pub fn repro<'a, T>()
where
    <View<&'a T> as Data>::Elem: Sized,
{
}

this breakage looks intended, in the sense that you'd get the same error on the old solver with only the Data bound without looking into the associated type

[lcnr]

fixed by also erroring here in the old solver rust-lang/rust#148477