New `Expected Option<Lifetime>, got &Option<Lifetime>` type mismatches in 1.61.0-beta.1

[lnicola]

crates/ide_assists/src/handlers/extract_function.rs 436:58-436:71: Expected Option<Lifetime>, got &Option<Lifetime>
crates/ide_assists/src/handlers/extract_function.rs 443:36-443:49: Expected Option<Lifetime>, got &Option<Lifetime>
crates/ide_assists/src/handlers/inline_call.rs 381:81-381:97: Expected Option<Type>, got &Option<Type>
crates/ide_assists/src/handlers/inline_call.rs 395:67-395:83: Expected GenericArgList, got &GenericArgList
crates/mbe/src/expander/matcher.rs 452:25-452:42: Expected Option<Separator>, got &Option<Separator>
crates/project_model/src/cargo_workspace.rs 337:28-337:46: Expected Option<String>, got &Option<String>

[lnicola]

Minimized:

fn foo() {
    match &Some(42) {
        s => bar(s.clone()), // expected `Option<i32>`, found `&Option<i32>`
    };
}

fn bar(s: Option<i32>) {}

[flodiebold]

(*s).clone() doesn't resolve at all; did something happen to the impl Clone for Option?

[flodiebold]

Do we support parsing the impl const syntax?

impl<T> const Clone for Option<T>
where
    T: ~const Clone + ~const Drop + ~const Destruct,

[Veykril]

We do

[flodiebold]

Then maybe it's the Destruct bound...

Edit:
Actually there's a

impl<T: ?Sized> const Destruct for T {}

though so I guess that's not it either...

I guess someone ™️ will have to copy the impls into a file and minimize them 😅

[lnicola]

We need an "inline type" assist 😄.

[flodiebold]

So, it's the ~const Drop bound, which we treat as just a Drop bound. i32 of course doesn't implement Drop. Apparently, if I understand correctly, ~const Drop had a different meaning than just "Drop during const eval". This has changed now, and the ~const Drop bound is removed on nightly, but we might have to special-case this just for 1.61 😕
On nightly though, the blanket impl for Destruct is gone, so we'll have to handle that instead. Kind of a moving target 😖