Rollup merge of #118396 - compiler-errors:ast-lang-items, r=cjgillot

Collect lang items from AST, get rid of `GenericBound::LangItemTrait`

r? `@cjgillot`
cc #115178

Looking forward, the work to remove `QPath::LangItem` will also be significantly more difficult, but I plan on doing it as well. Specifically, we have to change:
1. A lot of `rustc_ast_lowering` for things like expr `..`
2. A lot of astconv, since we actually instantiate lang and non-lang paths quite differently.
3. A ton of diagnostics and clippy lints that are special-cased via `QPath::LangItem`

Meanwhile, it was pretty easy to remove `GenericBound::LangItemTrait`, so I just did that here.

compiler/rustc_trait_selection/src/traits/error_reporting/suggestions.rs

@@ -4783,8 +4783,14 @@ pub fn suggest_desugaring_async_fn_to_impl_future_in_trait<'tcx>(
     };
 
     let future = tcx.hir_node_by_def_id(opaque_def_id).expect_item().expect_opaque_ty();
-    let Some(hir::GenericBound::LangItemTrait(_, _, _, generics)) = future.bounds.get(0) else {
-        // `async fn` should always lower to a lang item bound... but don't ICE.
+    let [hir::GenericBound::Trait(trait_ref, _)] = future.bounds else {
+        // `async fn` should always lower to a single bound... but don't ICE.
+        return None;
+    };
+    let Some(hir::PathSegment { args: Some(generics), .. }) =
+        trait_ref.trait_ref.path.segments.last()
+    else {
+        // desugaring to a single path segment for `Future<...>`.
         return None;
     };
     let Some(hir::TypeBindingKind::Equality { term: hir::Term::Ty(future_output_ty) }) =