Only suggest removing semicolon when expression is compatible with `impl Trait`

https://github.com/rust-lang/rust/issues/54771#issuecomment-476423690
> It still needs checking that the last statement's expr can actually conform to the trait, but the naïve behavior is there.

Only suggest removing a semicolon when the type behind the semicolon actually implements the trait in an RPIT `-> impl Trait`. Also upgrade the label that suggests removing the semicolon to a suggestion (should it be verbose?).

cc #54771

Better error for `for<...>` on associated type bound

With GATs just around the corner, we'll probably see more people trying out `Trait<for<'a> Assoc<'a> = ..>`.

This PR improves the syntax error slightly, and also makes it slightly easier to make this into real syntax in the future.

Feel free to push back if the reviewer thinks this should have a suggestion on how to fix it (i.e. push the `for<'a>` outside of the angle brackets), but that can also be handled in a follow-up PR.

`s/compiler-flags/compile-flags` in compiletest

Also make compiletest panic so this doesn't happen in the future! I literally always forget which it's called, so I wanted to make my life easier in the future.

Also open to the possibility of parsing both.

Rollup of 7 pull requests

Successful merges:

 - #95566 (Avoid duplication of doc comments in `std::char` constants and functions)
 - #95784 (Suggest replacing `typeof(...)` with an actual type)
 - #95807 (Suggest adding a local for vector to fix borrowck errors)
 - #95849 (Check for git submodules in non-git source tree.)
 - #95852 (Fix missing space in lossy provenance cast lint)
 - #95857 (Allow multiple derefs to be splitted in deref_separator)
 - #95868 (rustdoc: Reduce allocations in a `html::markdown` function)

Failed merges:

r? `@ghost`
`@rustbot` modify labels: rollup

rustdoc: Reduce allocations in a `html::markdown` function

Allow multiple derefs to be splitted in deref_separator

Previously in #95649 only a single deref within projection was supported and multiple derefs caused a bunch of issues, this PR fixes those issues.

```@oli-obk``` helped a ton again ❤️

Fix missing space in lossy provenance cast lint

See https://github.com/rust-lang/rust/pull/95599#discussion_r846425050

Check for git submodules in non-git source tree.

People occasionally download the source from https://github.com/rust-lang/rust/releases, but those source distributions will not work because they are missing the submodules. They will get a confusing `failed to load manifest for workspace member` error.

Unfortunately AFAIK there is no way to disable the GitHub source links. This change tries to detect this scenario and provide an error message that guides them toward a solution.

Closes #95608

Suggest adding a local for vector to fix borrowck errors

closes #95574

Suggest replacing `typeof(...)` with an actual type

This PR adds suggestion to replace `typeof(...)` with an actual type of `...`, for example in case of `typeof(1)` we suggest replacing it with `i32`.

If the expression
1. Is not const (`{ let a = 1; let _: typeof(a); }`)
2. Can't be found (`let _: typeof(this_variable_does_not_exist)`)
3. Or has non-suggestable type (closure, generator, error, etc)
we don't suggest anything.

The 1 one is sad, but it's not clear how to support non-consts expressions for `typeof`.

_This PR is inspired by [this tweet]._

[this tweet]: https://twitter.com/compiler_errors/status/1511945354752638976

Avoid duplication of doc comments in `std::char` constants and functions

For those consts and functions, only the summary is kept and a reference to the `char` associated const/method is included.

Additionaly, re-exported functions have been converted to function definitions that call the previously re-exported function. This makes it easier to add a deprecated attribute to these functions in the future.

Avoid accessing HIR from MIR passes

`hir_owner_nodes` contains a lot of information, and the query result is typically dirty. This forces dependent queries to be re-executed needlessly.

This PR refactors some accesses to HIR to go through more targeted queries that yield the same result.

Based on https://github.com/rust-lang/rust/pull/95435 and https://github.com/rust-lang/rust/pull/95436

update Miri

Fixes https://github.com/rust-lang/rust/issues/95844
r? `@ghost`

Make it possible to run `cargo test` for bootstrap

Note that this only runs bootstrap's self-tests, not compiler or library tests.

Helps with https://github.com/rust-lang/rust/issues/94829.

This allows to avoid looking at HIR from borrowck.

Remove ptr-int transmute in std::sync::mpsc

Since https://github.com/rust-lang/rust/pull/95340 landed, Miri with `-Zmiri-check-number-validity` produces an error on the test suites of some crates which implement concurrency tools<sup>*</sup>, because it seems like such crates tend to use `std::sync::mpsc` in their tests. This fixes the problem by storing pointer bytes in a pointer.

<sup>*</sup> I have so far seen errors in the test suites of `once_cell`, `parking_lot`, and `crossbeam-utils`.
(just updating the list for fun, idk)
Also `threadpool`, `async-lock`, `futures-timer`, `fragile`, `scoped_threadpool`, `procfs`, `slog-async`, `scheduled-thread-pool`, `tokio-threadpool`, `mac`, `futures-cpupool`, `ntest`, `actix`, `zbus`, `jsonrpc-client-transports`, `fail`, `libp2p-gossipsub`, `parity-send-wrapper`, `async-broadcast,` `libp2p-relay`, `http-client`, `mockito`, `simple-mutex`, `surf`, `pollster`, and `pulse`. Then I turned the bot off.

Fix `x doc compiler/rustc`

This also has a few cleanups to `doc.rs`. The last two commits I don't care about, but the first commit I'd like to keep - it will be very useful for https://github.com/rust-lang/rust/issues/44293.

Fixes https://github.com/rust-lang/rust/issues/95447.

Fix `cargo run` on Windows

Fixes the following error:
```
error: failed to run custom build command for `bootstrap v0.0.0 (C:\Users\Walther\git\rust\src\bootstrap)`

Caused by:
  process didn't exit successfully: `C:\Users\Walther\git\rust\target\debug\build\bootstrap-7757a4777dec0f86\build-script-build` (exit code: 101)
  --- stdout
  cargo:rerun-if-changed=build.rs
  cargo:rerun-if-env-changed=RUSTC
  cargo:rustc-env=BUILD_TRIPLE=x86_64-pc-windows-msvc
  cargo:rerun-if-env-changed=PATH

  --- stderr
  thread 'main' panicked at 'assertion failed: rustc.is_absolute()', src\bootstrap\build.rs:22:5
  note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
warning: build failed, waiting for other jobs to finish...
error: build failed
```

The problem was that the `dir.join` check only works with `rustc.exe`, not `rustc`.

Thanks `@Walther` for the help testing the fix!

Helps with https://github.com/rust-lang/rust/issues/94829.

Make def names and HIR names consistent.

The name in the `DefKey` is interned to create the `DefId`, so it does not
require any query to access.  This can be leveraged to avoid a few useless
HIR accesses for names.

~In order to achieve that, generic parameters created from universal
impl-trait are given the pretty-printed ast as a name, instead of
`{{opaque}}`.~

~Drive-by: the `TyCtxt::opt_item_name` used a dummy span for non-local
definitions.  We have access to `def_ident_span`, so we use it.~

Rollup of 7 pull requests

Successful merges:

 - #94794 (Clarify indexing into Strings)
 - #95361 (Make non-power-of-two alignments a validity error in `Layout`)
 - #95369 (Fix `x test src/librustdoc` with `download-rustc` enabled )
 - #95805 (Left overs of #95761)
 - #95808 (expand: Remove `ParseSess::missing_fragment_specifiers`)
 - #95817 (hide another #[allow] directive from a docs example)
 - #95831 (Use bitwise XOR in to_ascii_uppercase)

Failed merges:

r? `@ghost`
`@rustbot` modify labels: rollup

Note that this only runs bootstrap's self-tests, not compiler or library tests.

Use bitwise XOR in to_ascii_uppercase

This saves an instruction compared to the previous approach, which
was to unset the fifth bit with bitwise OR.

Comparison of generated assembly on x86: https://godbolt.org/z/GdfvdGs39

This can also affect autovectorization, saving SIMD instructions as well: https://godbolt.org/z/cnPcz75T9

Not sure if `u8::to_ascii_lowercase` should also be changed, since using bitwise OR for that function does not require an extra bitwise negate since the code is setting a bit rather than unsetting a bit. `char::to_ascii_uppercase` already uses XOR, so no change seems to be required there.

hide another #[allow] directive from a docs example

This is a repeat for Rc of e0e64a89,
which cleaned up the same thing for Arc.

expand: Remove `ParseSess::missing_fragment_specifiers`

It was used for deduplicating some errors for legacy code which are mostly deduplicated even without that, but at cost of global mutable state, which is not a good tradeoff.

cc https://github.com/rust-lang/rust/pull/95747#issuecomment-1091619403
r? ``@nnethercote``

Left overs of #95761

These are just nits. Feel free to close this PR if all modifications are not worth merging.

* `#![feature(decl_macro)]` is not needed anymore in `rustc_expand`
* `tuple_impls` does not require `$Tuple:ident`. I guess it is there to enhance readability?

r? ```@petrochenkov```

Fix `x test src/librustdoc` with `download-rustc` enabled

The problem was two-fold:
- Bootstrap was hard-coding that unit tests should always run with stage1, not stage2, and
- It hard-coded the sysroot layout in stage1, which puts libLLVM.so in `lib/rustlib/` instead of just `lib/`.

This also takes the liberty of fixing `test src/librustdoc --no-doc`, which has been broken since it was first added. It would be nice at some point to unify this logic with other tests; I opened a Zulip thread: https://rust-lang.zulipchat.com/#narrow/stream/131828-t-compiler/topic/Inconsistency.20in.20.60x.20test.60

Fixes https://github.com/rust-lang/rust/issues/91071.

Make non-power-of-two alignments a validity error in `Layout`

Inspired by the zulip conversation about how `Layout` should better enforce `size <= isize::MAX as usize`, this uses an N-variant enum on N-bit platforms to require at the validity level that the existing invariant of "must be a power of two" is upheld.

This was MIRI can catch it, and means there's a more-specific type for `Layout` to store than just `NonZeroUsize`.

It's left as `pub(crate)` here; a future PR could consider giving it a tracking issue for non-internal usage.