Currently, the rules which work with the tokens doesn't emit diagnostics after the location of the first syntax error. For example:

foo;

"hello world

bar;

This will only raise the useless-semicolon for foo and not bar because there's an unterminated string literal in between them.

Playground: https://play.ruff.rs/4c17a92c-0189-4b98-b961-27b04db14599

The task here is to disable this limit and allow token-based rules to check all the tokens. This raises a question: Now that the parser can now recover from an unclosed parenthesis (#11845), how to make sure that the rule logic knows about this and has the correct information about the nesting level? Should we reduce the nesting level if we encounter a Newline token?

We also need to make sure that this doesn't panic in any scenarios (valid or invalid source code). This can be done via lots of fuzzing.