End-to-end validation and production rollout

[gaiksaya]

Is your feature request related to a problem? Please describe

The release state machine spans 3 repos and multiple components (indices, Jenkins jobs, OSCAR agent, notifications). Without an end-to-end validation pass on a real release, we risk deploying something that fails at integration boundaries.

Describe the solution you'd like

• Register next upcoming release (e.g. 3.8.0) via the schedule job
• Run the release-state cron and verify state docs are indexed correctly
• Enable notifications to a test channel and verify escalating cadence
• Test ad-hoc queries and Go/No-Go decision capture
• Create OpenSearch Dashboards visualization (criteria timeline, R/Y/G history, decision audit trail)
• Update documentation (agents/release/README.md, DEVELOPER_GUIDE.md)
• Flip notification channel from test to production RM channel

Describe alternatives you've considered

• Deploying directly to production without dry-run — rejected because a bad notification or incorrect state could undermine RM trust in the system.

Acceptance Criteria

• Schedule registered: doc exists in opensearch_release_schedule for the test version
• State indexed: 12 criterion docs per version after cron run
• Notification received in test channel with correct format and cadence
• Ad-hoc query works: "What's the release state for X.Y.Z?" returns R/Y/G in Slack
• Decision capture works: Go/No-Go flow completes end-to-end
• Dashboard accessible with real data
• Documentation sufficient for a new contributor
• Production RM channel receives first real proactive notification