Rollup merge of #116219 - compiler-errors:relate-alias-ty-with-variance, r=lcnr

Relate alias ty with variance

In the new solver, turns out that the subst-relate branch of the alias-relate predicate was relating args invariantly even for opaques, which have variance 💀.

This change is a bit more invasive, but I'd rather not special-case it [here](aeaa5c30e5/compiler/rustc_trait_selection/src/solve/alias_relate.rs (L171-L190)) and then have it break elsewhere. I'm doing a perf run to see if the extra call to `def_kind` is that expensive, if it is, I'll reconsider.

r? ``@lcnr``

compiler/rustc_infer/src/infer/equate.rs

@@ -56,7 +56,7 @@ impl<'tcx> TypeRelation<'tcx> for Equate<'_, '_, 'tcx> {
         // performing trait matching (which then performs equality
         // unification).
 
-        relate::relate_args(self, a_arg, b_arg)
+        relate::relate_args_invariantly(self, a_arg, b_arg)
     }
 
     fn relate_with_variance<T: Relate<'tcx>>(

compiler/rustc_infer/src/infer/generalize.rs

@@ -183,7 +183,7 @@ where
             // Avoid fetching the variance if we are in an invariant
             // context; no need, and it can induce dependency cycles
             // (e.g., #41849).
-            relate::relate_args(self, a_subst, b_subst)
+            relate::relate_args_invariantly(self, a_subst, b_subst)
         } else {
             let tcx = self.tcx();
             let opt_variances = tcx.variances_of(item_def_id);