Port books to mdbook

Part of https://github.com/rust-lang/rust/issues/39588

blocked on https://github.com/rust-lang/rust/pull/39431

As a first step towards the bookshelf, we ~vendor mdbook in-tree and~ port our books to it. Eventually, both of these books will be moved out-of-tree, but the nightly book will rely on doing the same thing. As such, this intermediate step is useful.

r? @alexcrichton @brson

/cc @azerupi

Rollup of 8 pull requests

- Successful merges: #39659, #39730, #39754, #39772, #39785, #39788, #39790, #39813
- Failed merges:

Use check_variant for non_camel_case_types lint

This way we automatically consider lint attributes.

Fix #38452.

tidy: exempt URLs from the line length restriction

The length of a URL is usually not under our control, and Markdown
provides no way to split a URL in the middle.  Therefore, comment
lines consisting _solely_ of a URL (possibly with a Markdown link
label in front) should be exempt from the line-length restriction.

Inline hyperlink destinations ( `[foo](http://...)` notation ) are
_not_ exempt, because it is my arrogant opinion that long lines of
that type make the source text illegible.

The patch adds dependencies on the `regex` and `lazy_static` crates
to the tidy utility.  This _appears_ to Just Work, but if you would
rather not have that dependency I am willing to provide a hand-written
parser instead.

Add filename when running rustdoc --test on a markdown file

r? @alexcrichton

test: Remove sanitizer-thread test

Unfortunately it appears to spuriously fail so we can't gate on it

Adding compile fail test for staged_api feature

Issue #39059
r? @est31

@est31 running the tests for this feature fails. Is that expected since this is the `compile-fail`suite?

I copied this test from the run-pass suite: `rust/src/test/run-pass/reachable-unnameable-type-alias.rs`. What are the differences between these suites in operation and why they are used?

travis: Add builders without assertions

This commit adds three new builders, one OSX, one Linux, and one MSVC, which
will produce "nightlies" with LLVM assertions disabled. Currently all nightly
releases have LLVM assertions enabled to catch bugs before they reach the
beta/stable channels. The beta/stable channels, however, do not have LLVM
assertions enabled.

Unfortunately though projects like Servo are stuck on nightlies for the near
future at least and are also suffering very long compile times. The purpose of
this commit is to provide artifacts to these projects which are not distributed
through normal channels (e.g. rustup) but are provided for developers to use
locally if need be.

Logistically these builds will all be uploaded to `rustc-builds-alt` instead of
the `rustc-builds` folder of the `rust-lang-ci` bucket. These builds will stay
there forever (until cleaned out if necessary) and there are no plans to
integrate this with rustup and/or the official release process.

macros: fix ICE on certain sequence repetitions

Fixes #39709.
r? @nrc

Add equivalents of C's <ctype.h> functions to AsciiExt.

 * `is_ascii_alphabetic`
 * `is_ascii_uppercase`
 * `is_ascii_lowercase`
 * `is_ascii_alphanumeric`
 * `is_ascii_digit`
 * `is_ascii_hexdigit`
 * `is_ascii_punctuation`
 * `is_ascii_graphic`
 * `is_ascii_whitespace`
 * `is_ascii_control`

This addresses issue #39658.

Lightly tested on x86-64-linux.  tidy complains about the URLs in the documentation making lines too long, I don't know what to do about that.

Automate vendoring by invoking cargo-vendor when building src dist tarballs.

This avoids #39633 bringing the `src/vendor` checked into git by #37524, past 200,000 lines of code.

I believe the strategy of having rustbuild run `cargo vendor` during the `dist src` step is sound.

However, the only way to be sure `cargo-vendor` exists is to run `cargo install --force cargo-vendor`, which will recompile it every time (not passing `--force` means you can't tell between "already exists" and "build error"). ~~This is quite suboptimal and I'd like to somehow do it in each `Dockerfile` that would need it.~~

* [ ] Cache `CARGO_HOME` (i.e. `~/.cargo`) between CI runs
  * `bin/cargo-vendor` and the actual caches are the relevant bits
* [x] Do not build `cargo-vendor` all the time
  * ~~Maybe detect `~/.cargo/bin/cargo-vendor` already exists?~~
  * ~~Could also try to build it in a `Dockerfile` but do we have `cargo`/`rustc` there?~~
  * Final solution: check `cargo install --list` for a line starting with `cargo-vendor `

cc @rust-lang/tools

Add PartialOrd, Ord derivations to TypeId

I want to be able to sort a `Vec` of types which contain `TypeId`s, so an `Ord` derivation would be very useful to me. `Hash` and `PartialEq`/`Eq` already exist, so the missing `PartialOrd` and `Ord` derivations feel like an oversight to me.

Add intrinsics & target features for rd{rand,seed}

One question is whether or not we want to map feature name `rdrnd` to `rdrand` instead.

EDIT: as for use case, I would like to port my rdrand crate from inline assembly to these intrinsics.

Issue #39059
r? @est31

Rollup of 5 pull requests

- Successful merges: #39716, #39758, #39759, #39774, #39784
- Failed merges:

1. skip png files
2. skip fragments for the book and nomicon, as these are added by JS
3. Actually print the filename for errors

mdbook and rustdoc generate links differently, so we need to change all
these links.

We've decided that these deps are okay.

1. move everything under a src directory
2. add README.md to the SUMMARY.md

1. move everything under a src directory
2. add README.md to the SUMMARY.md

mdbook has a lot of optional dependencies that we don't want, so instead
of using it directly, we re-build rustbook to use mdbook as a library.
For convenience' sake, we keep the same CLI interface as mdbook; the
only difference is that it only accepts build and test subcommands,
rather than the full range.

The length of a URL is usually not under our control, and Markdown
provides no way to split a URL in the middle.  Therefore, comment
lines consisting _solely_ of a URL (possibly with a Markdown link
label in front) should be exempt from the line-length restriction.

Inline hyperlink destinations ( `[foo](http://...)` notation ) are
_not_ exempt, because it is my arrogant opinion that long lines of
that type make the source text illegible.

The patch adds dependencies on the `regex` and `lazy_static` crates
to the tidy utility.  This _appears_ to Just Work, but if you would
rather not have that dependency I am willing to provide a hand-written
parser instead.

typo fix

Remove '--disable-rustbuild' option from configure script.

Fixes https://github.com/rust-lang/rust/issues/39762.

add solaris rustbuild support

Add Solaris as recognized ostype
Add cputype recognition for Solaris

Fixes #39729

A future pull request will discriminate between the commercial release and older opensource derivatives to account for divergence, for now, this is compatible with both.

Remove duplicated "parameter" in E0089 text

Closes #39732

Add `swap` method for `Cell`

Addition to #39264

r? @alexcrichton

Unfortunately it appears to spuriously fail so we can't gate on it

They now always say how many lifetime / type parameters were expected
and are explicit about stating "lifetime" or "type" instead of just
"parameter".

Closes #39732

Fix some typos in the core::fmt docs.