Refactor signaling of E0223 to avoid multiple span_errs for same code.

(Doing this cuts out one diagnostic during the build.)

 Closes #21721.

 See https://github.com/rust-lang/rust/blob/41f8b1e89b5ca0c79d7bca782ca44085624d4564/src/libtest/lib.rs#L783-L788

 The branch \"else { continue }\" is the last code inside a loop body,
it is just useless.
Signed-off-by: NLai Jiangshan <laijs@cn.fujitsu.com>

 Changes .or() so that it can return a Result with a different E type
than the one it is called on.

Essentially:

    fn or(self, res: Result<T, E>) -> Result<T, E>

becomes

    fn or<F>(self, res: Result<T, F>) -> Result<T, F>

This brings `or` in line with the existing `and` & `or_else`

This is a
[breaking-change]
Due to some code needing additional type annotations.

 Change MarkerTrait to be invariant. This is a (small) loss of expressiveness, but is necessary for now to work around #22806. Fixes #22655.

r? @pnkfelix

 Keeps the method consistent with `Iterator::any`.

Closes #22617
[breaking-change]

 which starts happening after ~2 hours of machine uptime.
Closes #17845

 r? @steveklabnik

Closes #22698

I wasn't sure that this was appropriate for the book, but I've added this to the reference. I also noticed that one of the U+ symbols in the character literals section was missing the graves.

 r? @steveklabnik

 This patch contains a couple time fixes to make Rust compile on Bitrig again.  This does not affect OpenBSD.

   "body": null,

 Fixes #21370.

`unused-macro-with-follow-violation.rs` was already handled correctly. That test is just for good measure. :)

I have a more involved plan to clean this up, but it ran into difficulties such as #22814.

 Ensure we do not zero when \"moving\" types that are Copy.

Uses more precise `type_needs_drop` for deciding about emitting cleanup code.  Added notes about the weaknesses regarding `ty::type_contents` here.

Fix #22536

 We were recording stability attributes applied to fields in the
compiler, and even annotating it in the libs, but the compiler didn't
actually do the checks to give errors/warnings in user crates.

Details in the commit messages.

Implements `Debug`  for `RwLock` and `arc::Weak` in the same way it is implemented for `rc::Weak` (basically copy & paste).

The lack of this implementation prevents the automatic implementation of `Debug` for structs containing members of these types.

r? @alexcrichton
 
cc @nikomatsakis @pnkfelix 

Instead of allocating the same ports for ipv4 and ipv6 tests, instead draw all
ports from the same pool. Some tests connect to just "localhost" on a particular
port which may accidentally be interacting with other tests as the ipv-what-ness
isn't specified with the string "localhost"

Relevant logs:

* [Deadlock of the `net::tcp::tests::listen_localhost` test][mac]
* [Failure of the `fast_rebind` test][win1]
* [Failure of `multiple_connect_interleaved_lazy_schedule_ip4`][win2]

[mac]: https://gist.github.com/alexcrichton/349c7ce7c620c1adb2f2
[win1]: https://gist.github.com/alexcrichton/7e3611faae2e1edaee6f
[win2]: https://gist.github.com/alexcrichton/4f5f87749af3ad0f9851

Instead of allocating the same ports for ipv4 and ipv6 tests, instead draw all
ports from the same pool. Some tests connect to just "localhost" on a particular
port which may accidentally be interacting with other tests as the ipv-what-ness
isn't specified with the string "localhost"

Relevant logs:

* [Deadlock of the `net::tcp::tests::listen_localhost` test][mac]
* [Failure of the `fast_rebind` test][win1]
* [Failure of `multiple_connect_interleaved_lazy_schedule_ip4`][win2]

[mac]: https://gist.github.com/alexcrichton/349c7ce7c620c1adb2f2
[win1]: https://gist.github.com/alexcrichton/7e3611faae2e1edaee6f
[win2]: https://gist.github.com/alexcrichton/4f5f87749af3ad0f9851

rustdoc impl item did not include default methods for local crates, but did include them for external crates. This resulted in duplicate methods. Fix so that impl item does not include default methods for external crates.

Fix #22595.

Useful for debugging tests that hang forever.

unbreak openbsd/bitrig build after 2d200c9c (the complete build/check on openbsd is in progress)

@dhuseby maybe this one is in your list ?

In the process, also replaces two raw mutable pointers with `Unique` to
spell out the ownership semantics.

cc #22709

Closes #21721.

Fix #22536

(Instrumentation work for Issue 22536.)

The branch "else { continue }" is the last code inside a loop body,
it is just useless.
Signed-off-by: NLai Jiangshan <laijs@cn.fujitsu.com>

This needs to make it into the snapshot before #21824 can land.

r? @alexcrichton