Allow Clippy to define duplicate extra symbols

[Alexendoo]

Fixes #142410 (comment)

rustc_macros::symbols now also expands to a macro extra_symbols that calls rustc_macros::extra_symbols_impl with the prefill stream so that it can know which symbols are duplicated

extra_symbols allows supplying a name to define a macro to provide even more extra symbols, this is useful for external drivers that rely on clippy_utils e.g. TheBevyFlock/bevy_cli#456

There's a usage example in compiler/rustc_span/src/symbol/tests.rs

DUPLICATE_SYMBOLS is currently unused, but we can use it on the clippy side to remove duplicates without interfering with r-l/rust PRs

[rustbot]

r? @fee1-dead

rustbot has assigned @fee1-dead.
They will have a look at your PR within the next two weeks and either review your PR or reassign to another reviewer.

Use r? to explicitly pick a reviewer

[rustbot]

Some changes occurred in src/tools/clippy

cc @rust-lang/clippy

[flip1995]

Clippy changes LGTM

[fee1-dead]

This seems to be quite above my personal threshold for the compiler to maintain and support this, since it isn't a mechanism used by the compiler itself. Do you think there could be an easier solution here? For example, rustc may expose the number of symbols, then clippy could scan the rustc list (either the file itself, or exposed through a constant) to error at compile time if there is a duplicate symbol.

Or clippy could use a LazyLock.

[bors]

☔ The latest upstream changes (presumably #144449) made this pull request unmergeable. Please resolve the merge conflicts.

[Alexendoo]

An update on this:

clippy could scan the rustc list (either the file itself, or exposed through a constant) to error at compile time if there is a duplicate symbol.

This works with the new const PartialEq but currently it's a too slow to do the linear scans. With const PartialOrd from #143800 we could do a binary search so I'll mark this as blocked for now

In the meanwhile #146347 improves the runtime error