test: enforce F811 so a duplicate definition cannot silently replace the first

[ryan-crabbe-berri]

TLDR

Problem this solves:

• A name bound twice keeps only the second binding
• A local can shadow an import the module still calls
• 344 such sites, and nothing stops the next one

How it solves it:

• Selects ruff F811 in ruff-tests.toml, already wired to CI
• 311 were repeated imports and came out with ruff's own fix
• 33 needed a decision: shadowed imports, rebound closures, fixtures
• Load-test fixtures move to a conftest, which is how pytest shares them

User Flow

Before: a developer edits the watsonx credential lookup and the test that checks which secrets it reads can stop compiling its own assertion

1. They set WATSONX_API_KEY and send POST https://litellm-domain/v1/chat/completions for a watsonx model
2. The call succeeds, because the gateway read the key from the environment
3. Someone changes which environment names that lookup tries, and the test guarding it builds its expected list with a variable named call that has quietly replaced the call helper the same file imports
4. The next person to use that helper in the file gets a confusing failure that has nothing to do with their change, because the name no longer means what the import says it means

After: the shadowing is rejected before it lands

1. The same edit is made
2. CI flags the local that took over an imported name, so the file keeps one meaning per name
3. The same check refuses a second def test_x that would silently replace the first, so a test cannot disappear by being renamed onto its neighbour

Relevant issues

Linear ticket

Pre-Submission checklist

• I have added meaningful tests
• The handful of test files covering my change pass locally
• My PR passes all required CI/CD checks (e.g., lint, schema.d.ts sync check, etc.)
• My PR's scope is as isolated as possible; it only solves 1 specific problem
• I have received a Greptile Confidence Score of at least 4/5 before requesting a maintainer review

Screenshots / Proof of Fix

This PR changes lint config and tests, so there is no proxy route to curl. The check that matters for a change made mostly of deletions is that no test went missing and no module stopped importing, so both sides collect the whole suite and the counts are compared.

Before (4b29702)

1. Run

pytest tests --collect-only -q --continue-on-collection-errors

2. Output:

46989/46991 tests collected (2 deselected), 70 errors in 20.12s

After (e9d40a8)

1. Run

pytest tests --collect-only -q --continue-on-collection-errors

2. Output:

46989/46991 tests collected (2 deselected), 70 errors in 41.76s

3. Run

ruff check --config ruff-tests.toml tests

4. Output:

All checks passed!

Rule and regression numbers

ruff check --config ruff-tests.toml tests went from 344 F811 findings to zero. Identical collection on both sides: the same 46989 tests, and the same 70 pre-existing collection errors, all of them modules that fail to import for missing credentials or optional dependencies. Nothing new and nothing gone.

Type

🧹 Refactoring
🚄 Infrastructure
✅ Test

Caveats (if any)

• Nine prisma_client parameters keep a narrow noqa with the reason
• tests/load_tests/conftest.py is new, so pytest shares those fixtures
• 311 of 344 sites are ruff's own safe fix, so the diff is mostly deletions

Final Attestation

• The tests check the right things, including the edge cases, and regressions in the respective real-world customer use-cases are not possible after this PR

[greptile-apps]

Too many files changed for review (172 files, 100 file limit).

Bypass the limit by tagging @greptile-apps to review.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!