Like #43008 (f6689991), but _much more aggressive_.

Add method `String::retain`

Behaves like `Vec::retain`, accepting a predicate `FnMut(char) -> bool`
and reducing the string to only characters for which the predicate
returns `true`.

Add Redox Dockerfile and Travis Environment

This adds Redox to the Travis build. This is an example implementation of https://github.com/rust-lang/rust/issues/43206

Ship the rustdoc book

Fixes #42322, as it's the last step.

Blocked on https://github.com/rust-lang/rust/pull/43790, though they will not conflict.

r? @rust-lang/docs

emit StorageLive for box temporaries

We started emitting StorageDead, so we better emit the corrseponding
StorageLive to avoid problems.

cc #43772 solson/miri#303

Behaves like `Vec::retain`, accepting a predicate `FnMut(char) -> bool`
and reducing the string to only characters for which the predicate
returns `true`.

Rollup of 6 pull requests

- Successful merges: #43756, #43790, #43846, #43848, #43862, #43868
- Failed merges:

we don't want to support plugins

Add missing newline in Deref docs to fix rendering

Fixes #43866.

(Verified locally.)

r? @steveklabnik

rustdoc: put auto-hidden docblock labels in line with the toggle

before:

![image](https://user-images.githubusercontent.com/5217170/29279951-0cf21d86-80df-11e7-86d5-bfd76c5df429.png)

after:

![image](https://user-images.githubusercontent.com/5217170/29279962-18479ce2-80df-11e7-952a-d22cb03965e5.png)

(images taken from `std::heap::AllocErr`)

Rewrite/reorganize docs for stack size/thread names for spawned threads.

* Moves docs about stack size and thread naming from `Builder` to the
  `std::thread` module
* Adds more links to the new module-level documentation
* Mentions the 2 MiB stack size default, but indicate it's subject to
  change

Fixes https://github.com/rust-lang/rust/issues/43805.

Remove my mailmap entry.

I no longer want this mailmap entry for myself.

Write the "passes" chapter of the rustdoc book

cc #42322

r? @rust-lang/docs

Instant is monotonically nondecreasing

We don't want to guarantee that `Instant::now() != Instant::now()` is
always true since that depends on the speed of the processor and the
resolution of the clock.

remove the "defaulted unit" type bit during writeback

The defaulted unit bit is only relevant for the surrounding inference
context, and can cause trouble, including spurious lints and ICEs,
outside of it.

Fixes #43853.

r? @EddyB

Fixes #43866.

Fix "Mis-calculated spans" errors from `-Z save-analysis` + refactoring

Removed the path span extraction methods from `SpanUtils`:

* spans_with_brackets
* spans_for_path_segments
* spans_for_ty_params

Use the `span` fields in `PathSegment` and `TyParam` instead.

(Note that since it processes `ast::Path` not a qualified path (`hir::QPath` / `ast::QSelf`), UFCS path will be flattened: `<Foo as a::b::c::Trait>::D::E::F::g` will be seen as `a::b::c::Trait::D::E::F::g`.)

Fix #43796. Close #41478.

r? @nrc

cc #42322

Implement `RefCell::replace` and `RefCell::swap`

Tracking issue: #43570

Use hir::ItemLocalId as keys in TypeckTables.

This PR makes `TypeckTables` use `ItemLocalId` instead of `NodeId` as key. This is needed for incremental compilation -- for stable hashing and for being able to persist and reload these tables. The PR implements the most important part of https://github.com/rust-lang/rust/issues/40303.

Some notes on the implementation:
* The PR adds the `HirId` to HIR nodes where needed (`Expr`, `Local`, `Block`, `Pat`) which obviates the need to store a `NodeId -> HirId` mapping in crate metadata. Thanks @EddyB for the suggestion! In the future the `HirId` should completely replace the `NodeId` in HIR nodes.
* Before something is read or stored in one of the various `TypeckTables` subtables, the entry's key is validated via the new `TypeckTables::validate_hir_id()` method. This makes sure that we are not mixing information from different items in a single table.

That last part could be made a bit nicer by either (a) new-typing the table-key and making `validate_hir_id()` the only way to convert a `HirId` to the new-typed key, or (b) just encapsulate sub-table access a little better. This PR, however, contents itself with not making things significantly worse.

Also, there's quite a bit of switching around between `NodeId`, `HirId`, and `DefIndex`. These conversions are cheap except for `HirId -> NodeId`, so if the valued reviewer finds such an instance in a performance critical place, please let me know.

Ideally we convert more and more code from `NodeId` to `HirId` in the future so that there are no more `NodeId`s after HIR lowering anywhere. Then the amount of switching should be minimal again.

r? @EddyB, maybe?

Mark closures return via impl-trait as reachable.

This should fix some of the open `impl trait` issues, like #40839, #43135, and #35870.

r? @EddyB

We started emitting StorageDead, so we better emit the corrseponding
StorageLive to avoid problems.

The defaulted unit bit is only relevant for the surrounding inference
context, and can cause trouble, including spurious lints and ICEs,
outside of it.

Fixes #43853.

Use the `span` field in PathSegment and TyParam instead.

Fix #43796. Close #41478.

Don't inline debug methods

The inner methods aren't inlined, so this puts more pressure on LLVM for
literally no benefit.

Closes #43843

ast_validation: forbid "nonstandard" literal patterns

Since #42886, macros can create "nonstandard" PatKind::Lit patterns,
that contain path expressions instead of the usual literal expr. These
can cause trouble, including ICEs.

We *could* map these nonstandard patterns to PatKind::Path patterns
during HIR lowering, but that would be much effort for little gain, and
I think is too risky for beta. So let's just forbid them during AST
validation.

Fixes #43250.

beta-nominating because regression.
r? @EddyB

Cleanup for "Support compiling rustc without LLVM (try 2)"

This includes a small patch to allow running tests without llvm. Also check if you are not trying to compile a dylib.

cc #42932
r? @alexcrichton

The inner methods aren't inlined, so this puts more pressure on LLVM for
literally no benefit.

Closes #43843