fix #48816 don't print help on indirect compiler ICE

https://github.com/rust-lang/rust/issues/48816#issuecomment-372817573

Speed up SVH computation by using Fingerprint::combine()

Fix #47297

Clarify usage message for --remap-path-prefix.

Add a test for #48070

Resolves #48070.
The bug itself was fixed by #48770, but that PR didn't add a test for it.

r? @nikomatsakis

rustc: Don't invoke `lld` with an `@`-file

Looks like LLD doesn't support this yet, so always try to use the OS before we
fall back to using `@`

cc https://github.com/rust-lang/rust/issues/48948

Move librustc_typeck READMEs to rustc guide

cc https://github.com/rust-lang-nursery/rustc-guide/issues/2 and #48478

Don't merge this before https://github.com/rust-lang-nursery/rustc-guide/pull/85

r? @nikomatsakis

Add missing examples

r? @QuietMisdreavus

tweak code fences in the rustdoc book

You can stack backticks to create "big code fences" if you're documenting some markdown and need to have code fences inside your code fences. This is especially important in this spot in the Rustdoc Book, because we're showing that using no language specifier on your code blocks is interpreted as using `rust`, but the code blocks here lose their code fences!

`````````markdown
``````markdown
Just showing some recursion, nbd.

```rust
println!("sup");
```
``````

(If you have edit powers in the rust-lang/rust repo, hit the edit button to view the source and see even more recursion :P)
`````````

Remove some unnecessary IdxSet methods

This replaces `IdxSet:: reset_to_empty` with `IdxSet:: clear`, and `IdxSet::elems`/`IdxSet::each_bit` with `IdxSet::iter`. Based on some [comments on #rustc](https://botbot.me/mozilla/rustc/2018-01-23/?msg=96063396).

r? @pnkfelix

Fix blink when main theme is selected

r? @QuietMisdreavus

Add info message for -Wall command

Users coming from other languages (namely C and C++) often expect
to use a -Wall flag. Rustc doesn't support that, and previously it
simply printed that it didn't recognize the "all" lint.

This change makes rustc print out a help message, explaining:
- Why there is no -Wall flag
- How to view all the available warnings
- Point out that the most commonly used warning is -Wunused
- Instead of using a command-line flag, the user should consider
  a !#[warn(unused)] directive in the root of their crate.

I tried to keep the language consistent with the other usage help. Comment if I should change anything.

closes #10234, if accepted.

Remove syntax and syntax_pos thread locals

This moves `syntax` and `syntax_pos` globals into a struct which are pointed to by thread locals. Most of the changes here are indentation changes in test. It would probably be a good idea to ignore whitespace changes while reviewing. Some indentation is unchanged to avoid merge conflicts.

r? @michaelwoerister

Update libc to 0.2.39

CC #42681

r? @alexcrichton

Put back ui json check

r? @petrochenkov

Fix #47297

r? @michaelwoerister

Move ascii::escape_default to libcore

As requested in #46409, the `ascii::escape_default` method has been added to the core library. All I did was copy over the `std::ascii` module file, remove the (redundant) `AsciiExt` trait, and change some of the documentation to match. None of the tests were changed.

I wasn't sure how to handle the annotations. For `EscapeDefault` and `escape_default()`, I changed them to `#[unstable(feature = "core_ascii", issue = "46409")]`. Is that alright? Or should I leave them as they were?

Resolves #48070.
The bug itself was fixed by #48770, but that PR didn't add a test for it.

introduce canonical queries, use for normalization and dropck-outlives

This branch adds in the concept of a **canonicalized trait query** and uses it for three specific operations:

- `infcx.at(cause, param_env).normalize(type_foldable)`
    - normalizes all associated types in `type_foldable`
- `tcx.normalize_erasing_regions(param_env, type_foldable)`
    - like normalize, but erases regions first and in the result; this leads to better caching
- `infcx.at(cause, param_env).dropck_outlives(ty)`
    - produces the set of types that must be live when a value of type `ty` is dropped
    - used from dropck but also NLL outlives

This is a kind of "first step" towards a more Chalk-ified approach. It leads to a **big** speedup for NLL, which is basically dominated by the dropck-outlives computation. Here are some timing measurements for the `syn` crate (pre-branch measurements coming soon):

| Commit | NLL disabled | NLL enabled |
| ------- | --- | --- |
| Before my branch | 5.43s | 8.99s |
| After my branch | 5.36s | 7.25s |

(Note that NLL enabled still does *all the work* that NLL disabled does, so this is not really a way to compare the performance of NLL versus the AST-based borrow checker directly.) Since this affects all codepaths, I'd like to do a full perf run before we land anything.

Also, this is not the "final point" for canonicalization etc. I think canonicalization can be made substantially faster, for one thing. But it seems like a reasonable starting point for a branch that's gotten a bit larger than I would have liked.

**Commit convention:** First of all, this entire branch ought to be a "pure refactoring", I believe, not changing anything about external behavior. Second, I've tagged the most important commits with `[VIC]` (very important commit), so you can scan for those. =)

r? @EddyB

Consolidate `trans_apply_param_substs` and
`trans_apply_param_substs_env`. Also remove `trans_impl_self_ty`

Also, perform substitution in smaller parts.

Fixes #31567
Fixes #47470
Fixes #48132
Fixes #48179

In particular:

- `fully_normalize_monormophic_ty` => `normalize_erasing_regions`
- `normalize_associated_type_in_env` => `normalize_erasing_regions`
- `fully_normalize_associated_types_in` => `normalize_erasing_regions`
- `erase_late_bound_regions_and_normalize` => `normalize_erasing_late_bound_regions`

Backed by a canonicalized query. This computes all the types/regions that need
to be live when the destructor runs (i.e., that the dtor may access).

It is backed by the new `normalize_projection_ty` query, which uses
canonicalization.

Before, the identifier `X` was also used when generating a pattern
to match against the dep-node. So `Foo(DefId)` would generate a match
pattern like:

    match foo {
        Foo(DefId) => ...
    }

This does not scale to more general types like `&'tcx
Ty<'tcx>`. Therefore, we now require *exactly one* argument (the macro
was internally tupling anyway, and no actual nodes use more than one
argument), and then we can generate a fixed pattern like:

    match foo {
        Foo(arg) => ...
    }

Huzzah. (Also, hygiene is nice.)