Auto merge of #120942 - compiler-errors:deep-assoc-hang, r=lcnr

Ignore own item bounds in parent alias types in `for_each_item_bound`

Fixes #120912

I want to get a vibe check on this approach, which is very obviously a hack, but I believe something that is forwards-compatible with a more thorough solution and "good enough for now".

The problem here is that for a really deep rigid associated type, we are now repeatedly considering unrelated item bounds from the parent alias types, meaning we're doing a *lot* of extra work in the MIR inliner for deeply substituted rigid projections.

This feels intimately related to #107614. In that PR, we split *supertrait* bounds (bound which share the same `Self` type as the predicate which is being elaborated) and *implied* bounds (anything that is implied by elaborating the predicate).

The problem here is related to the fact that we don't maintain the split between these two for `item_bounds`. If we did, then when recursing into a parent alias type, we'd want to consider only the bounds that are given by [`PredicateFilter::All`](https://doc.rust-lang.org/nightly/nightly-rustc/rustc_hir_analysis/astconv/enum.PredicateFilter.html#variant.SelfOnly) **except** those given by [`PredicateFilter::SelfOnly`](https://doc.rust-lang.org/nightly/nightly-rustc/rustc_hir_analysis/astconv/enum.PredicateFilter.html#variant.SelfOnly).

compiler/rustc_trait_selection/src/traits/select/mod.rs

@@ -1605,6 +1605,8 @@ impl<'cx, 'tcx> SelectionContext<'cx, 'tcx> {
         on_ambiguity: impl FnOnce(),
     ) -> ControlFlow<T, ()> {
         let mut idx = 0;
+        let mut in_parent_alias_type = false;
+
         loop {
             let (kind, alias_ty) = match *self_ty.kind() {
                 ty::Alias(kind @ (ty::Projection | ty::Opaque), alias_ty) => (kind, alias_ty),
@@ -1618,6 +1620,18 @@ impl<'cx, 'tcx> SelectionContext<'cx, 'tcx> {
             for bound in
                 self.tcx().item_bounds(alias_ty.def_id).instantiate(self.tcx(), alias_ty.args)
             {
+                // HACK: On subsequent recursions, we only care about bounds that don't
+                // share the same type as `self_ty`. This is because for truly rigid
+                // projections, we will never be able to equate, e.g. `<T as Tr>::A`
+                // with `<<T as Tr>::A as Tr>::A`.
+                if in_parent_alias_type {
+                    match bound.kind().skip_binder() {
+                        ty::ClauseKind::Trait(tr) if tr.self_ty() == self_ty => continue,
+                        ty::ClauseKind::Projection(p) if p.self_ty() == self_ty => continue,
+                        _ => {}
+                    }
+                }
+
                 for_each(self, bound, idx)?;
                 idx += 1;
             }
@@ -1627,6 +1641,8 @@ impl<'cx, 'tcx> SelectionContext<'cx, 'tcx> {
             } else {
                 return ControlFlow::Continue(());
             }
+
+            in_parent_alias_type = true;
         }
     }