Caching the stable hash of Ty within itself

Instead of computing stable hashes on types as needed, we compute it during interning.

This way we can, when a hash is requested, just hash that hash, which is significantly faster than traversing the type itself.

We only do this for incremental for now, as incremental is the only frequent user of stable hashing.

As a next step we can try out

* moving the hash and TypeFlags to Interner, so projections and regions get the same benefit (tho regions are not nested, so maybe that's not a good idea? Would be nice for dedup tho)
* start comparing types via their stable hash instead of their address?

Rollup of 6 pull requests

Successful merges:

 - #92399 (fix typo in btree/vec doc: Self -> self)
 - #92823 (Tweak diagnostics)
 - #94248 (Fix ICE when passing block to while-loop condition)
 - #94414 (Fix ICE when using Box<T, A> with large A)
 - #94445 (4 - Make more use of `let_chains`)
 - #94449 (Add long explanation for E0726)

Failed merges:

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

Add long explanation for E0726

This is the cleaned up version of #87655 with the missing fixes.

Part of https://github.com/rust-lang/rust/issues/61137.

r? `@Urgau`

4 - Make more use of `let_chains`

Continuation of #94376.

cc #53667

Fix ICE when using Box<T, A> with large A

A sequel to #94043 that fixes #81270 and #92054 (duplicate).

Fix ICE when passing block to while-loop condition

We were incorrectly delaying a bug when we passed _any_ block (that evaluated to `()`) to a while loop. This PR makes the check a bit more sophisticated.

We should only suppress the error here in cases that are equivalent to those we find in #93574 (i.e. only while loop conditions that have destructuring assignment expressions in them).

Fixes #93997
cc `@estebank` who added this code

I would not be opposed to removing the delay-bug altogether, and just emitting this error always. I much prefer duplicate errors over no errors.

Tweak diagnostics

* Recover from invalid `'label: ` before block.
* Make suggestion to enclose statements in a block multipart.
* Point at `match`, `while`, `loop` and `unsafe` keywords when failing
  to parse their expression. (Fix #92705.)
* Do not suggest `{ ; }`.
* Do not suggest `|` when very unlikely to be what was wanted (in `let`
  statements).

fix typo in btree/vec doc: Self -> self

this pr fixes #92345
the documentation refers to the object the method is called for, not the type, so it should be using the lower case self.

* Recover from invalid `'label: ` before block.
* Make suggestion to enclose statements in a block multipart.
* Point at `match`, `while`, `loop` and `unsafe` keywords when failing
  to parse their expression.
* Do not suggest `{ ; }`.
* Do not suggest `|` when very unlikely to be what was wanted (in `let`
  statements).

Rollup of 5 pull requests

Successful merges:

 - #89793 (Add `slice::{from_ptr_range, from_mut_ptr_range} `)
 - #92642 (Update search location from a relative path to absolute)
 - #93389 (regression for issue 90847)
 - #93413 (Fix broken link from rustdoc docs to ayu theme)
 - #94365 (Fix MinGW target detection in raw-dylib)

Failed merges:

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

Fix MinGW target detection in raw-dylib

LLVM target doesn't have to be the same as Rust target so relying on it is wrong.

It was one of concerns in https://github.com/rust-lang/rust/pull/88801 that was not fixed in https://github.com/rust-lang/rust/pull/90782.

Fix broken link from rustdoc docs to ayu theme

regression for issue 90847

Adds a regression test for issue #90847

Update search location from a relative path to absolute

This should address issue #90311.

Add `slice::{from_ptr_range, from_mut_ptr_range} `

Adds `slice::{from_ptr_range, from_mut_ptr_range}` as counterparts to `slice::{as_ptr_range, as_mut_ptr_range}`.

more complete sparc64 ABI fix for aggregates with floating point members

Previous fix didn't handle nested structures at all.

Continuation of #94376.

cc #53667

Only create a single expansion for each inline integration.

The inlining integrator used to create one expansion for each span from the callee body.
This PR reverses the logic to create a single expansion for the whole call,
which is more consistent with how macro expansions work for macros.

This should remove the large memory regression in #91743.

Apply noundef metadata to loads of types that do not permit raw init

This matches the noundef attributes we apply on arguments/return types.

Fixes (partially) #74378.

update Miri

Fixes https://github.com/rust-lang/rust/issues/94318
r? `@ghost`

Rollup of 7 pull requests

Successful merges:

 - #94396 (1 - Make more use of `let_chains`)
 - #94397 (Document that pre-expansion lint passes are softly deprecated)
 - #94399 (Add test for #79465 to prevent regression)
 - #94409 (avoid rebuilding bootstrap when PATH changes)
 - #94415 (Use the first codegen backend in the config.toml as default)
 - #94417 (Fix duplicated impl links)
 - #94420 (3 - Make more use of `let_chains`)

Failed merges:

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

Apply noundef attribute to all scalar types which do not permit raw init

Beyond `&`/`&mut`/`Box`, this covers `char`, enum discriminants, `NonZero*`, etc.
All such types currently cause a Miri error if left uninitialized,
and an `invalid_value` lint in cases like `mem::uninitialized::<char>()`.

Note that this _does not_ change whether or not it is UB for `u64` (or
other integer types with no invalid values) to be undef.

Fixes (partially) #74378.

r? `@ghost` (blocked on #94127)

`@rustbot` label S-blocked

3 - Make more use of `let_chains`

Continuation of #94376.

cc #53667

Fix duplicated impl links

Fixes #78701.

The problem is that the blanket impl has the same ID as the other impl, except that we don't derive IDs when we generate the sidebar. We now do.

r? ``@notriddle``

Use the first codegen backend in the config.toml as default

It is currently hard coded to llvm if enabled and cranelift otherwise.
This made some sense when cranelift was the only alternative codegen
backend. Since the introduction of the gcc backend this doesn't make
much sense anymore. Before this PR bootstrapping rustc using a backend
other than llvm or cranelift required changing the source of
rustc_interface. With this PR it becomes a matter of putting the right
backend as first enabled backend in config.toml.

cc ```@antoyo```

avoid rebuilding bootstrap when PATH changes

Fixes https://github.com/rust-lang/rust/issues/94408

r? ```@Mark-Simulacrum```

Add test for #79465 to prevent regression

Fixes #79465.

Like this we will be able to close the issue.

r? ````@matthiaskrgr````

Document that pre-expansion lint passes are softly deprecated

The pre-expansion lint pass has been softly deprecated since https://github.com/rust-lang/rust/pull/69838. Every once in a while I see someone mention it as a possibility, only get the feedback that it's deprecated. This PR officially documents that the method is soft deprecated to have a single point of truth for it.

That's it. Have a great rest of the day 🙃

---

* See [rust#69838](https://github.com/rust-lang/rust/pull/69838)
* See [rust-clippy#5518](https://github.com/rust-lang/rust-clippy/pull/5518)

1 - Make more use of `let_chains`

Continuation of #94376.

cc #53667

For MIRI, cfg out the swap vectorization logic from 94212

Because of #69488 the swap logic from #94212 doesn't currently work in MIRI.

Copying in smaller pieces is probably much worse for its performance anyway, so it'd probably rather just use the simple path regardless.

Part of #94371, though another PR will be needed for the CTFE aspect.

r? `@oli-obk`
cc `@RalfJung`

This matches the noundef attributes we apply on arguments/return types.