The starting point for this was identical comments on two different
fields, in `ast::VariantData::Struct` and `hir::VariantData::Struct`:
```
    // FIXME: investigate making this a `Option<ErrorGuaranteed>`
    recovered: bool
```
I tried that, and then found that I needed to add an `ErrorGuaranteed`
to `Recovered::Yes`. Then I ended up using `Recovered` instead of
`Option<ErrorGuaranteed>` for these two places and elsewhere, which
required moving `ErrorGuaranteed` from `rustc_parse` to `rustc_ast`.

This makes things more consistent, because `Recovered` is used in more
places, and there are fewer uses of `bool` and
`Option<ErrorGuaranteed>`. And safer, because it's difficult/impossible
to set `recovered` to `Recovered::Yes` without having emitted an error.

Make a minimal amount of region APIs public

Tools like Creusot, Prusti or Gillian-Rust need to access information about the loans and regions that exist in MIR programs. While `rustc` provides information about loans, there is currently no public way to reason about the regions present in a MIR program. In particular, we to know which regions are actually equal to each other and which ones outlive each other. Currently, `rustc` provides access to `RegionInferenceContext` but the public api hides that last portion of the information.

This PR proposes to make a few apis public, allowing verifiers to reason about the lifetimes present in Rust programs:
- [eval_equal](https://doc.rust-lang.org/beta/nightly-rustc/rustc_borrowck/region_infer/struct.RegionInferenceContext.html#method.eval_equal)
- [eval_outlives](https://doc.rust-lang.org/beta/nightly-rustc/rustc_borrowck/region_infer/struct.RegionInferenceContext.html#method.eval_outlives)
- (Optional) [constraint_sccs](https://doc.rust-lang.org/beta/nightly-rustc/rustc_borrowck/region_infer/struct.RegionInferenceContext.html#method.constraint_sccs)

The first two functions would allow us to compare regions and from this we can construct the set of `RegionVid` which are actually equal to each other, and then recover the inclusions between those regions, while the second allows for more direct, but _low level_ access to that information.

…ranteed, r=compiler-errors

Add `ErrorGuaranteed` to `Recovered::Yes` and use it more.

The starting point for this was identical comments on two different fields, in `ast::VariantData::Struct` and `hir::VariantData::Struct`:
```
    // FIXME: investigate making this a `Option<ErrorGuaranteed>`
    recovered: bool
```
I tried that, and then found that I needed to add an `ErrorGuaranteed` to `Recovered::Yes`. Then I ended up using `Recovered` instead of `Option<ErrorGuaranteed>` for these two places and elsewhere, which required moving `ErrorGuaranteed` from `rustc_parse` to `rustc_ast`.

This makes things more consistent, because `Recovered` is used in more places, and there are fewer uses of `bool` and
`Option<ErrorGuaranteed>`. And safer, because it's difficult/impossible to set `recovered` to `Recovered::Yes` without having emitted an error.

r? `@oli-obk`

…piler-errors

interpret/miri: better errors on failing offset_from

Fixes rust-lang/miri#3104

chore: remove repetitive words

… r=est31

Make `#![feature]` suggestion MaybeIncorrect

Fixes rust-lang/rust-clippy#12784

The `unstable_name_collisions` lint uses `disabled_nightly_features` to mention the feature name, but accepting the suggestion would result in an ambiguity error

There are other calls where accepting the feature gate would fix code when ran with `cargo fix --broken-code`, though it's not always desirable to add a feature gate even if the user is currently on nightly so MaybeIncorrect seems appropriate