Fix unwrap_expect_used in test modules defined by complex configuration predicates (#16369)

[corentinguilloteau]

This PR fixes detection of test contexts in complex scenarios (e.g., #[cfg(all(test, ...))]). This allows the unwrap_used and expect_used lints to correctly suppress warnings in test contexts with compound cfg conditions.

Note that this also updates the behavior of lints that indirectly depend on is_cfg_test (around 20 lints).

Fixes #16369

changelog: [unwrap_expect_used]: Fix in test modules defined by complex configuration predicates

[rustbot]

r? @llogiq

rustbot has assigned @llogiq.
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

Why was this reviewer chosen?

The reviewer was selected based on:

• Owners of files modified in this PR: 7 candidates
• 7 candidates expanded to 7 candidates
• Random selection from Jarcho, llogiq, samueltardieu

[samueltardieu]

Hi Corentin, thanks for your contribution.

It would be great if you could handle the CfgEntry::Not case as well, or at least clearly state in the function documentation why it is not handled. There is a risk that some other code in Clippy wants to use cfg_implies_test() for the reverse purpose (ensure that we cannot be in a test settings), and this function would return None for #[cfg(not(test))] while the caller could expect it to return Some(false).

r? samueltardieu
@rustbot author

View changes since this review

[rustbot]

Reminder, once the PR becomes ready for a review, use @rustbot ready.

[corentinguilloteau]

I've reworked the initial implementation with a new approach that handles more configuration predicates than the old one (most notably, it now handles the CfgEntry::Not case).
Solving the general problem would likely require a SAT solver, which feels like unnecessary complexity here. This new algorithm remains comparable in complexity to the previous one while covering a vast majority of real-world predicates.

I've also improved the API so that using cfg_implies_test_is() for the reverse use case becomes less ambiguous.

@rustbot ready