feat(db): add automatic purge of expired terminated clusters

[r2dedios]

Summary

• Adds a purge_expired_clusters() SQL function that deletes terminated clusters older than a configurable retention period (default: 365 days) and refreshes materialized views
• Adds a daily pg_cron job (0 0 * * *) to run the purge automatically
• Adds CIQ_MAX_DATA_AGE env var to configure retention in both Compose and Helm deployments

Details

Terminated clusters and all their dependent data (instances, tags, expenses, events, schedules) accumulate indefinitely. This adds a database-level maintenance mechanism to purge stale data without exposing a destructive API endpoint.

The cascade chain handles cleanup automatically:

DELETE cluster → instances (CASCADE) → tags, expenses, target_instances (CASCADE)
                                     → events (trigger)
              → target_clusters (CASCADE) → schedules → action_runs (CASCADE)
              → events (trigger)

Files changed

File	Change
db/sql/init.sql	purge_expired_clusters() function
db/sql/cron.sql	Daily pg_cron job
deployments/compose/compose-devel.yaml	CIQ_MAX_DATA_AGE env var + override step
deployments/helm/cluster-iq/values.yaml	database.maxDataAge setting
deployments/helm/cluster-iq/templates/database/job.yaml	Env var + override step
deployments/helm/cluster-iq/templates/database/configmap-init.yaml	SQL function + pg_cron job

Test plan

• Deployed locally with CIQ_MAX_DATA_AGE=1 and verified SELECT purge_expired_clusters(1); removes terminated clusters
• Verified cascade deletes dependent data (instances, tags, expenses, events)
• Verified materialized views refresh after purge
• Verified pg_cron job is registered via SELECT * FROM pg_cron_history();