"is greater or equal to". Beside common sense.

library/backtrace also declares a feature called `gimli-symbolize` which
appear used, but the feature in std with the same name is unused, so
remove it.

`ConstKind::Value` is the only variant where control flow leaves the
first match on `impl_ct.kind()`, so there is no need for a second match
on the same expression later on.

Sort them, and remove the unused ones (`lint_reasons` and
`proc_macro_span`).

They each have a single call site.

Control flow never gets past the end of the `ExpandResult::Retry` match
arm, due to the `span_bug` and the `continue`. Therefore, the code after
the match can only be reached from the `ExpandResult::Ready` arm.

This commit moves that code after the match into the
`ExpandResult::Ready` arm, avoiding the need for the `continue` in the
`ExpandResult::Retry` arm.

Too clever by half, IMO.

Alongside the existing `Invocation::span`.

Because the type is not obvious, and this clarifies things.

It is currently an enum and the `tts` and `idx` fields are repeated
across the two variants.

This commit splits it into a struct `Frame` and an enum `FrameKind`, to
factor out the duplication. The commit also renames `Frame::new` as
`Frame::new_delimited` and adds `Frame::new_sequence`. I.e. both
variants now have a constructor.

…ompiler-errors

deref patterns: impl `DerefPure` for more std types

Context: [deref patterns](rust-lang#87121). The requirements of `DerefPure` aren't precise yet, but these types unambiguously satisfy them.

Interestingly, a hypothetical `impl DerefMut for Cow` that does a `Clone` would *not* be eligible for `DerefPure` if we allow mixing deref patterns with normal patterns. If the following is exhaustive then the `DerefMut` would cause UB:
```rust
match &mut Cow::Borrowed(&()) {
    Cow::Owned(_) => ..., // Doesn't match
    deref!(_x) if false => ..., // Causes the variant to switch to `Owned`
    Cow::Borrowed(_) => ..., // Doesn't match
    // We reach unreachable
}
```

io safety: update Unix explanation to use `Arc`

Fixes rust-lang#124384

Cc ```@jsgf```

…s, r=Amanieu

String.truncate comment microfix (greater or equal)

String.truncate calls Vec.truncate, in turn, and that states "is greater or equal to". Beside common sense.

run-make-support: preserve tooks.mk behavior for EXTRACXXFLAGS

In rust-lang#123149 when trying to add a command wrapper for `cc`, I didn't preserve the behavior of tools.mk completely: tools.mk had

```makefile
# Extra flags needed to compile a working executable with the standard library
ifdef IS_WINDOWS
ifdef IS_MSVC
	#EXTRACFLAGS := ws2_32.lib userenv.lib advapi32.lib bcrypt.lib ntdll.lib synchronization.lib
else
	#EXTRACFLAGS := -lws2_32 -luserenv -lbcrypt -lntdll -lsynchronization
	EXTRACXXFLAGS := -lstdc++
	#EXTRARSCXXFLAGS := -l static:-bundle=stdc++
endif
else
ifeq ($(UNAME),Darwin)
	#EXTRACFLAGS := -lresolv
	EXTRACXXFLAGS := -lc++
	#EXTRARSCXXFLAGS := -lc++
else
ifeq ($(UNAME),FreeBSD)
	#EXTRACFLAGS := -lm -lpthread -lgcc_s
else
ifeq ($(UNAME),SunOS)
	#EXTRACFLAGS := -lm -lpthread -lposix4 -lsocket -lresolv
else
ifeq ($(UNAME),OpenBSD)
	#EXTRACFLAGS := -lm -lpthread -lc++abi
	#RUSTC := $(RUSTC) -C linker="$(word 1,$(CC:ccache=))"
else
	#EXTRACFLAGS := -lm -lrt -ldl -lpthread
	EXTRACXXFLAGS := -lstdc++
	#EXTRARSCXXFLAGS := -lstdc++
endif
endif
endif
endif
endif
```

Note that for {`FreeBSD`, `SunOs`, `OpenBSD`} the `-lstdc++` flag is *not* passed, so `EXTRACXXFLAGS` for those platforms should be an empty `vec![]`.

r? ghost (testing this with PR CI)

…kingjubilee

library/std: Remove unused `gimli-symbolize` feature

library/backtrace also declares a feature called `gimli-symbolize` which appear used, but the feature in std with the same name is unused, so remove it.

… r=compiler-errors

`rustc_expand` cleanups

Some cleanups I made while looking through this code. Nothing that requires any real domain-specific knowledge about this crate.

r? ````@michaelwoerister````

variable-precision float operations can differ depending on optimization levels

Follow-up to rust-lang#121793 and rust-lang#118217 that accounts for optimizations changing the precision of these functions.

Fixes rust-lang#109118
Fixes rust-lang#71355

Tweak `consts_may_unify`

r? ````@lcnr````

const_eval_select: add tracking issue

…yn, r=compiler-errors

AST pretty: Use `builtin_syntax` for type ascription

Follow-up to rust-lang#122806.
CC #124619.