[red-knot] Infer precise types for len() calls

[InSyncWithFoo]

Summary

Resolves #14598.

Test Plan

Markdown tests.

[github-actions]

ruff-ecosystem results

Linter (stable)

✅ ecosystem check detected no linter changes.

Linter (preview)

✅ ecosystem check detected no linter changes.

[carljm]

Thank you!! A few comments.

[AlexWaygood]

Thanks!

[carljm]

This looks quite close! Thanks for your patience with all the review comments.

I think it makes sense to leave the diagnostics as TODO here, because it will be a lot easier to emit these diagnostics once #14760 lands (we won't have to return some indicator such that inference emits the diagnostic, we'll be able to just emit it directly where we find the issue.)

[MichaReiser]

@carljm I'm not sure if #14760 will land. It's unclear to me how suppressions would work with accumulators whereas it's clear to me how they would work in our existing model.

[carljm]

@MichaReiser Ok, let's discuss those details elsewhere. The only part that's relevant here is to say that I think accumulators will be super useful in type checking and make a lot of things much simpler and easier, so I think we really want/need it, even if it requires more work on handling suppressions.

Looks like all comments in this review were addressed.

[carljm]

Thanks! And great find on the string-literal-unpacking issue. Merging.