Change orders and exports that get stuck now unstick themselves.
Change order validation runs that stall now auto-fail. If a validation run sits in "in progress" for more than 30 minutes, it gets marked as failed with a clear error message. No more phantom runs blocking your workflow.
Export jobs that hang get cleaned up automatically. Jobs stuck in limbo for over an hour now fail gracefully with a readable error. The system checks every 5 minutes.
Permanently failed exports require explicit acknowledgment to retry. If an export hit a permanent failure — bad schema, missing data, something structural — you can't just hit retry anymore. You have to acknowledge what went wrong first. Stops the same export from failing the same way on loop.
