Safely shut down cronjobs

When Payload is configured to automatically create cron jobs (e.g., via config.jobs.autoRun), our CI began experiencing frequent MongoDB and SQLite connection errors. This stemmed from two core issues:

• Integration tests are executed in a single process across all test suites. If one suite schedules cron jobs, those jobs continue running even as the next test suite starts - even if that suite has no autoRun jobs configured.
• When Jest shuts down, it tears down the in-memory database in the same process as Payload. If a cron job runs during this shutdown and performs a DB operation, it can trigger a race condition, leading to flaky test failures.

Fixes:

• MongoMemoryServer shutdown is now moved to a separate process, ensuring that teardown happens cleanly after the Payload/Jest process exits.
• Introduces a new payload.destroy() method, which gracefully shuts down all registered cron jobs before calling payload.db.destroy(). This ensures no cron jobs persist across test suites, as payload.destroy() is now invoked in the afterAll() hook of each suite.

Docs improvement

Fixes #12658

Clarifies that jobs.autoRun only runs existing jobs already in the queue - it does not enqueue jobs itself.

Adds e2e test coverage

Adds an e2e test to verify jobs.autoRun functionality, which previously had no test coverage.