Diagnose and fix a flaky test

Help me find out why a test is flaky and fix it properly. Test code: [PASTE THE TEST] Code under test: [PASTE THE RELEVANT PARTS] Failure output: [PASTE ONE OR MORE FAILING RUNS] How often it fails: [e.g. "1 in 20 runs", "only in CI", "only when run with the full suite"] Test environment: [FRAMEWORK, CI RUNNER, PARALLELISM, DATABASE OR CONTAINERS USED] Work through the common causes and say which fit my evidence: - Timing: sleeps, timeouts, async work not awaited, animations, polling. - Order dependence: shared state, leftover database rows, global mocks, caches between tests. - Concurrency: parallel tests sharing ports, files, or records. - Time and randomness: current date, time zones, daylight saving, unseeded random data, unordered collections. - Environment: network calls, external services, resource limits in CI. Return: 1. The most likely cause(s), ranked, with the evidence for each. 2. How to reproduce it reliably (e.g. run in a loop, fixed seed, specific test order, throttled CPU). 3. The fix, as code. Prefer fixing the root cause: deterministic waits, isolated data, injected clocks, seeded randomness. 4. If the real bug is in the production code rather than the test, say so clearly. 5. What not to do: explain why a blanket retry or longer sleep would only hide the problem.