Keep Failures Independent

When one job feeds several channels, like an alert that goes to email and to Slack, a failure in one shouldn't block or repeat the others. Say that each channel succeeds or fails on its own, and how a retry should avoid sending anything twice. Vague: "Email and Slack the on-call team about each urgent ticket." Specific: "Email and Slack the on-call team about each urgent ticket. Track each channel separately: if email is down, Slack still goes out once, and email retries later without re-posting to Slack. Simulate an email outage and show me what each channel did."