fix(tests): resolve flaky integration tests caused by slug collisions

[iulspop]

Summary

• Make factory-generated organization slugs unique by default (append cuid suffix) to prevent collisions from faker's limited company name pool
• Make happy-path test names unique with createId() so route-produced slugs don't collide with stale data
• Properly await global setup's ensureStripeProductsAndPricesExist() instead of fire-and-forget

Closes #127

Test plan

• Integration tests pass 5 consecutive full-suite runs (242/242 tests each)
• Verify no regressions in unit tests and happy-dom tests

🤖 Generated with Claude Code

[janhesters]

@iulspop thank you so much for this PR 🙌

I really appreciate you digging into the flaky tests.

That said, I think we should investigate this a bit more deeply and see if we can arrive at a more elegant fix.

My gut feeling is that the underlying issue might be in our test setup rather than the individual test cases. Instead of working around it in the tests, I’d prefer to understand what’s actually causing the collisions.

A couple of thoughts:

1. Regarding slug collisions:
   If we’re seeing slug collisions, that suggests our middleware isn’t doing what it’s supposed to. We have logic that should prevent collisions (see here). If that’s failing, that’s something we should address directly.

2. Regarding name collisions:
   Did you notice these collisions happening within the same test run, or only across different runs?

   • If it’s across runs, that could point to issues in our cleanup logic. Ideally, all integration tests with Vitest should use onTestFinished to clean up any data they create. Our Playwright E2E tests also usually have explicit teardown logic at the end.
   • If it’s within the same run, then maybe the right fix is to tweak createPopulatedOrganization() so it appends the id to the organization name by default, rather than adjusting individual tests.

Would you be willing to take a closer look and help us figure out what’s really going on here?

[iulspop]

Would you be willing to take a closer look and help us figure out what’s really going on here?

Yes, I will and get back to you. ETA a week.

[janhesters]

@iulspop Sounds good, thank you!