[pyflakes] Flag annotated variable redeclarations as F811 in preview mode

[RenzoMXD]

Summary

• In preview mode, F811 now detects annotated variable redeclarations like bar: int = 1; bar: int = 2 as unused redefinitions
• Plain reassignments (x = 1; x = 2) remain unflagged, as this is normal Python
• Only triggers when both the original and the shadowing binding are annotated assignments with values
  • bar: int = 1; bar: int = 2 — flagged
  • bar = 1; bar = 2 — not flagged (plain reassignment)
  • bar = 1; bar: int = 2 — not flagged (mixed)
  • bar: int = 1; bar = 2 — not flagged (mixed)
  • bar: int = 1; print(bar); bar: int = 2 — not flagged (first is used)

Closes #23802

Test plan

• Added F811_34.py test fixture covering all combinations (annotated/plain, used/unused)
• All 35 existing F811 tests pass with no regressions
• Full pyflakes suite passes (463 tests)

[astral-sh-bot]

ruff-ecosystem results

Linter (stable)

✅ ecosystem check detected no linter changes.

Linter (preview)

ℹ️ ecosystem check detected linter changes. (+1 -0 violations, +0 -0 fixes in 1 projects; 55 projects unchanged)

prefecthq/prefect (+1 -0 violations, +0 -0 fixes)

ruff check --no-cache --exit-zero --no-fix --output-format concise --preview

+ src/prefect/client/schemas/responses.py:366:5: F811 Redefinition of unused `concurrency_options` from line 359: `concurrency_options` redefined here

Changes by rule (1 rules affected)

code	total	+ violation	- violation	+ fix	- fix
F811	1	1	0	0	0

[dylwil3]

Thanks for your interest in contributing to Ruff! Could you please take a look at the AI policy and make any necessary adjustments?

[RenzoMXD]

@dylwil3 I checked every line one by one and all are good implemented. I used a BindingFlags flag instead of checking the statement directly. The preview gate is in the F811 checker rather than in redefines() because the semantic crate doesn't have access to linter settings. I updated the function in mod.rs. I think my PR is good but I expect any review from you.

[dylwil3]

Ok, thank you! I think we can get away with something a bit simpler without introducing a new flag. I pushed a different approach, we'll see if CI passes but it ought to be equivalent.

[dylwil3]

Thanks! Just need some documentation.