Implement feature(precise_capturing_of_types)

[compiler-errors]

1. Set uncaptured ty/ct substs as bivariant. I'm almost certain is still quite broken, and I'll come up with a follow-up PR hopefully soon exercising some of the ways that it doesn't work today (related to the fact that "invariant xform bivariant = invariant").
2. Validate that uncaptured ty/ct substs don't show up in the hidden type. I also reworked uncaptured ty/ct errors for TAITs. That diagnostic could also be improved tbh.
3. Add a couple simple tests.

r? lcnr or reassign if you're not interested in reviewing feature work

tracking:

• Tracking issue for RFC 3617 precise capturing of types #130043

[lcnr]

how does this avoid unconstrained type variables when doing something like

fn foo<T>() -> impl Sized + use<> {}

fn main() {
    let x = foo<u32>(); // opaque<u32>;
    let y = x; // opaque<?unconstrained>
}

[compiler-errors]

it doesn't afaict :> this feature is still pretty broken, that's why it's incomplete

[bors]

☔ The latest upstream changes (presumably #138464) made this pull request unmergeable. Please resolve the merge conflicts.

[lcnr]

also extract this into a subfn?

[lcnr]

it doesn't afaict :> this feature is still pretty broken, that's why it's incomplete

do you have a plan for how to fix this? I feel like the feature as is doesn't have much value and worry that we likely can't stabilize it without large changes

however, I don't think the cost of supporting this is meaningfully large, so r=me after nits