Don't use hir_ty_to_ty in result_large_err

[Jarcho]

fixes #9414

This occurs starting with 2022-09-01. I checked that this does fix the ICE on rust-lang/rust@9353538. Not sure which pr caused the late-bound region to leak through hir_ty_to_ty.

changelog: None

[rust-highfive]

r? @Alexendoo

(rust-highfive has picked a reviewer for you, use r? to override)

[Alexendoo]

Ah damn, I thought about suggesting to replace the hir_ty_to_ty in #9373 but left it as it wasn't a new addition

Thanks!

@bors r+

[bors]

📌 Commit bd70ccf has been approved by Alexendoo

It is now in the queue for this repository.

[bors]

⌛ Testing commit bd70ccf with merge 958a9cf...

[bors]

☀️ Test successful - checks-action_dev_test, checks-action_remark_test, checks-action_test
Approved by: Alexendoo
Pushing 958a9cf to master...

[lukaslueg]

@Jarcho, @Alexendoo If you could, maybe you have two hints about what had been wrong in the original PR?

[Jarcho]

The error is from a change on rustc's side. hir_ty_to_ty isn't erasing the lifetime bounds on trait objects in some cases so you're left with a type like for<'a> (dyn 'a + Trait) and such a type can't be used when computing layouts. The old type would have been (dyn 'erased Trait) (not real syntax) which is a way of saying to ignore the lifetime value as it doesn't matter.

[flip1995]

The original lint implementation didn't hit beta yet and this was already synced, so no need to backport. Thanks for labeling it anyway!