default auto traits: use default supertraits instead of Self: Trait bounds on associated items

[Bryanskiy]

First commit: the motivation has been discussed here.

Second commit: the only new places where new implicit DefaultAutoTrait bounds are generated are supertraits and trait object so ?Trait syntax should be extended to these places only.

r? @lcnr

[rustbot]

HIR ty lowering was modified

cc @fmease

[fmease]

Please remove the above FIXME

[fmease]

Moreover this test is quite a mixed bag, it's unclear what the test intentions are. In #142693 I almost deleted it for that reason but I kept it to track bugs pertaining to more_maybe_bounds instead (as can be seen by the top-level FIXME).

I don't think we should test "this relaxed bound is not permitted here" in this file, I'd advise you to create a new one whose name is based on tests/ui/unsized/relaxed-bounds-invalid-places.rs (obv not inside unsized/; maybe have both test files in trait-bounds/ actually).

[Bryanskiy]

From my understanding there are 2 checks for relaxed bounds consistency:

• Syntactic check during AST->HIR lowering (validate_relaxed_bound) to ensure that ?Trait bounds are in a right place.
• Semantic check during HIR->ty lowering (check_and_report_invalid_relaxed_bounds) to ensure that ? modifier is applied to "right" traits. And this check has to be done regardless of the place and results of the first check.

This PR fixes the first check but not the second and therefore FIXME comments in this file are still relevant.

So, your suggestion is to keep these checks in separate files for readability reasons, right? In my opinion it isn't very useful because the invocation of second check implies that the first check has already been invoked. That is, the test that covers the second check also covers the first one.

[rustbot]

This PR was rebased onto a different master commit. Here's a range-diff highlighting what actually changed.

Rebasing is a normal part of keeping PRs up to date, so no action is needed—this note is just to help reviewers.