List open backup alerts.

Returns all alerts that have not yet been resolved, ordered by opened_at descending (newest first). The UI renders these as a banner above the Backups page content. Alerts are auto-opened by the watcher and auto-resolved when the triggering condition clears.

Schedule overdue — the backup scheduler did not enqueue a job within the expected window (1 hour past next_run). Usually means the scheduler task is dead or wedged.

Job stalled — a backup_jobs row has been in state='pending' for more than 1 hour. The runner never claimed the job. Usually means the runner task is dead or the runner concurrency cap is too low.