#127 CRITICAL: workflow_run.status not updated when task permanently fails

During disruption testing, tasks exhaust max_attempts and go to 'failed' state, but workflow_run.status remains 'pending'. Evidence: - workflow_run.status = 'pending' - absurd.t_*.state = 'failed' - absurd.r_*.state = 'failed' (all retries exhausted) Root cause: fail_run() does not update workflow_run table when task permanently fails. Impact: - Workflows appear stuck in 'pending' forever - No recovery possible without manual intervention - Tests timeout waiting for completion Fix: Update workflow_run.status to 'failed' when task exhausts retries in fail_run() or orchestrator.