It was only using the `infcx` to "accumulate" constraints anyhow.

All recursive calls only use type-vars that should already be
resolved, because they were obtained from `ty`.

We want any add'l context required to be passed through the struct
itself.

Rollup of 11 pull requests

Successful merges:

 - #51104 (add `dyn ` to display of dynamic (trait) types)
 - #51153 (Link panic and compile_error docs)
 - #51642 (Fix unknown windows build)
 - #51730 (New safe associated functions for PinMut)
 - #51731 (Fix ICEs when using continue as an array length inside closures (inside loop conditions))
 - #51747 (Add error for using null characters in #[export_name])
 - #51769 (Update broken rustc-guide links)
 - #51786 (Remove unnecessary stat64 pointer casts)
 - #51788 (Fix typo)
 - #51789 (Don't ICE when performing `lower_pattern_unadjusted` on a `TyError`)
 - #51791 (Minify css)

Failed merges:

r? @ghost

Minify css

Sizes changes:

```
dark.css: 8821 => 7804 (~11%)
light.css: 8587 => 7565 (~11%)
rustdoc.css: 22364 => 17818 (~20%)
settings.css: 1384 => 1236 (~10%)
```

So obviously, the bigger the file, the bigger minification.

r? @QuietMisdreavus

Don't ICE when performing `lower_pattern_unadjusted` on a `TyError`

Fix #50577. CC #51696.

r? @oli-obk

Fix typo

Remove unnecessary stat64 pointer casts

In effect, these just casted `&mut stat64` to `*mut stat64`, twice.
That's harmless, but it masked a problem when this was copied to new
code calling `fstatat`, which takes a pointer to `struct stat`.  That
will be fixed by #51785, but let's remove the unnecessary casts here
too.

Update broken rustc-guide links

Recently, there has been some rearrangement of the content in the Rustc
Guide, and this commit changes the urls the match the updated guide.

Add error for using null characters in #[export_name]

Fixes #51741.

Fix ICEs when using continue as an array length inside closures (inside loop conditions)

Fixes #51707.
Fixes #51708.

r? @estebank

New safe associated functions for PinMut

- Add safe `get_mut` and `map`
- Rename unsafe equivalents to `get_mut_unchecked` and `map_unchecked`

The discussion about this starts [in this comment](https://github.com/rust-lang/rust/issues/49150#issuecomment-399604573) on the tracking issue.

Fix unknown windows build

Fixes #51618.

Link panic and compile_error docs

This adds documentation links between `panic!()` and `compile_error!()` as per #47275, which points out that they’re similar. It also adds a sentence to the `compile_error()` docs I thought could be added.

add `dyn ` to display of dynamic (trait) types

~~I'm not sure we want the `dyn` in the E0277 "trait bound [...] is not satisfied" messages ("bound" sounds like a different thing in contrast to the names of specific trait-object types like `Box<dyn Trait>`), but I'm finding the code I would need to change that hard to follow—the [display object seems to](https://github.com/rust-lang/rust/blob/f0805a4421449bd6fe3096d63820fbebe2bfcd1d/src/librustc/traits/error_reporting.rs#L600) be a [`Predicate::Trait`](https://github.com/rust-lang/rust/blob/f0805a4421449bd6fe3096d63820fbebe2bfcd1d/src/librustc/ty/mod.rs#L962) variant, whose [`Display` implementation](https://github.com/rust-lang/rust/blob/f0805a4421449bd6fe3096d63820fbebe2bfcd1d/src/librustc/util/ppaux.rs#L1309) calls `.print` on its `PolyTraitPredicate` member, [which is a type alias](https://github.com/rust-lang/rust/blob/f0805a4421449bd6fe3096d63820fbebe2bfcd1d/src/librustc/ty/mod.rs#L1112) for `ty::Binder<TraitPredicate<'tcx>>`, whose [`Display` implementation](https://github.com/rust-lang/rust/blob/f0805a4421449bd6fe3096d63820fbebe2bfcd1d/src/librustc/util/ppaux.rs#L975-L985) ... _&c._— so maybe it's time to pull-request this and see what reviewers think.~~

 Resolves #49277 (?).

r? @nikomatsakis

Implementation of RFC 2086 - Allow Irrefutable Let patterns

This is the set of changes for RFC2086. Tracking issue #44495. Rendered [here](https://github.com/rust-lang/rfcs/pull/2086)

Obligation forest cleanup

While looking at this code I was scratching my head about whether a node could appear in both `parent` and `dependents`. Turns out it can, but it's not useful to do so, so this PR cleans things up so it's no longer possible.

Fix possibly endless loop in ReadDir iterator

Certain directories in `/proc` can cause the `ReadDir` iterator to loop indefinitely. We get an error code (22) when calling libc's `readdir_r` on these directories, but `entry_ptr` is `NULL` at the same time, signalling the end of the directory stream.

This change introduces an internal state to the iterator such that the `Some(Err(..))` value will only be returned once when calling `next`. Subsequent calls will return `None`.

fixes #50619

`Self` in where clauses may not be object safe

Needs crater, virtually certain to cause regressions.

In #50781 it was discovered that our object safety rules are not sound because we allow `Self` in where clauses without restrain. This PR is a direct fix to the rules so that we disallow methods with unsound where clauses.

This currently uses hard error to measure impact, but we will want to downgrade it to a future compat error.

Part of #50781.

r? @nikomatsakis

Use fstatat64 where available

None

In effect, these just casted `&mut stat64` to `*mut stat64`, twice.
That's harmless, but it masked a problem when this was copied to new
code calling `fstatat`, which takes a pointer to `struct stat`.  That
will be fixed by #51785, but let's remove the unnecessary casts here
too.

build: add llvm-tools to manifest

This commit expands on a previous commit to build llvm-tools as a rustup component. It causes the llvm-tools component to be built if the extended step is active. It also adds llvm-tools to the build manifest so rustup can find it.

I tested this as far as I could, but had to hack `build-manifest/src/main.rs` a bit as it is not supported on MacOS. The main change I am not sure about is this line:

```rust
self.package("llvm-tools", &mut manifest.pkg, TARGETS);
```

There are numerous calls to `self.package()`, and I'm not sure if `TARGETS`, `HOSTS`, or `["*"]` is appropriate for llvm-tools.

Otherwise I mostly copied the example set by `rustfmt-preview`.

three diagnostics upgrades

 * reword `...` expression syntax error to not imply that you should use it in patterns either (#51043) and make it a structured suggestion
 * shorten the top-line message for the trivial-casts lint by tucking the advisory sentence into a help note
 * structured suggestion for pattern-named-the-same-as-variant warning

r? @oli-obk