Side-step ICE on RePlaceholder in eval_verify_bound

[estebank]

Address #76168.

I don't think this is the correct way of handling this, but it does get rid of a stable-to-stable regression ICE.

The previous output (before the regression) was:

error[E0310]: the parameter type `F` may not live long enough
  --> src/lib.rs:9:1
   |
9  | / {
10 | | }
   | |_^
   |
   = help: consider adding an explicit lifetime bound `F: 'static`...

while the output with this PR is

error: `<F as FnOnce<(&'a mut i32,)>>::Output` does not live long enough
  --> $DIR/issue-76168.rs:8:1
   |
LL | / {
LL | |     let mut i = 41;
LL | |     f(&mut i).await;
LL | | }
   | |_^

[rust-highfive]

r? @petrochenkov

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

[petrochenkov]

r? @rust-lang/borrowck

[matthewjasper]

I had a look at this, and I think that the correct fix is to replace the placeholder regions in the type here

rust/compiler/rustc_borrowck/src/type_check/constraint_conversion.rs

Lines 98 to 111 in 40ebd07

	GenericArgKind::Type(t1) => {
	// we don't actually use this for anything, but
	// the `TypeOutlives` code needs an origin.
	let origin = infer::RelateParamBound(DUMMY_SP, t1, None);
	TypeOutlives::new(
	&mut *self,
	tcx,
	region_bound_pairs,
	implicit_region_bound,
	param_env,
	)
	.type_must_outlive(origin, t1, r2);
	}

Do you want to update this PR to do that, or should I open a new PR with that change?

[estebank]

@matthewjasper would you have the time to open a PR doing the change you propose? I'm a bit swamped at the moment. You can close this PR if you do so. I'll keep this one open for now just as a note to self, to remind me to look at this if there's no movement in the medium term.

[matthewjasper]

Opened #91321

[crlf0710]

Triage: Since #91321 has landed and #76168.is now closed, i'm closing this PR.