Add regression tests for #55219 and #55241

Also another test where a duplicate-like error appears to have been
suppressed; I'm not 100% sure why this output changes, though I could
imagine that some duplicate suppression is enabled by this PR.

I confirmed that it fails on:
rustdoc 1.30.0-beta.12 (96a22982 2018-10-04)
and passes on:
rustdoc 1.31.0-nightly (f99911a4 2018-10-23)

This UI test is sensitive to backtrace output, so it should make sure
that backtraces are not enabled by the environment.

Fix rustdoc and fulldeps tests

This fixes `issue-28848.rs` -- it also handles another case that the
AST region checker gets wrong (`wf-self-type.rs`).  I don't actually
think that this is the *right way* to be enforcing this constraint --
I think we should probably do it more generally, perhaps by editing
`predicates_of` for the impl itself. The chalk-style implied bounds
setup ought to fix this.

This reverts commit 3cc8f738.

Co-Authored-By: Nestebank <esteban@kuber.com.ar>

Suggest using `'static` when a lifetime is missing in the return type
with a structured suggestion instead of a note.

In particular, after the first for a given region variable. This
suppresses a lot of duplicate errors.

Fixes #54124

Felix S. Klock II pointed out that this suggestion (introduced in
pull-request #43178 / eac74104) was being issued for one-field-struct
expected types (in which case it is misleading and outright wrong),
even though it was only intended for one-field enum-variants (most
notably, `Some`). Particularly tender-hearted code-historians may be
inclined to show mercy towards the author of #43178 on the grounds
that it's somewhat confusing that struct field definitions are given
in a type called `ty::VariantDef`.

Add a conditional to adhere to the original intent. (It would be
possible to generalize to structs, but not obviously net desirable.)
This adds a level of indentation, so the diff here is going to be
easier to read in ignore-whitespace mode (`-w`).

Resolves #55250.

This gives at least some explanation for why a borrow is expected to
last for a certain free region. Also:

* Reports E0373: "closure may outlive the current function" with NLL.
* Special cases the case of returning a reference to (or value
  referencing) a local variable or temporary (E0515).
* Special case assigning a reference to a local variable in a closure
  to a captured variable.

Enable one fully ignored test

Except those testing parsing during macro expansion

It would be kind of embarrassing to ship with the "issue TBD" message!

But its test was written in an outdated way that hits a different error despite IHLE, so keep a variant around for that case.

It's already stable in 2018; this finishes the stabilization.

In order to output a path that could actually be imported (valid and
visible), we need to handle re-exports correctly.

For example, take `std::os::unix::process::CommandExt`, this trait is
actually defined at `std::sys::unix::ext::process::CommandExt` (at time
of writing).

`std::os::unix` rexports the contents of `std::sys::unix::ext`.
`std::sys` is private so the "true" path to `CommandExt` isn't accessible.

In this case, the visible parent map will look something like this:

(child) -> (parent)
`std::sys::unix::ext::process::CommandExt` -> `std::sys::unix::ext::process`
`std::sys::unix::ext::process` -> `std::sys::unix::ext`
`std::sys::unix::ext` -> `std::os`

This is correct, as the visible parent of `std::sys::unix::ext` is in fact
`std::os`.

When printing the path to `CommandExt` and looking at the current
segment that corresponds to `std::sys::unix::ext`, we would normally
print `ext` and then go to the parent - resulting in a mangled path like
`std::os::ext::process::CommandExt`.

Instead, we must detect that there was a re-export and instead print `unix`
(which is the name `std::sys::unix::ext` was re-exported as in `std::os`).