Parameterize monitoring data retention by plan

Summary

• parameterize VictoriaMetrics retention through VM_RETENTION_PERIOD
• pass QUERYID_RETENTION_HOURS into the monitoring Flask backend
• document defaults and suggested short/long retention values in .env.example
• add CI coverage for default, override, bare-integer, long-retention, and empty-fallback Docker Compose render behavior
• preserve existing VictoriaMetrics safety flags, including -search.maxQueryDuration=30s
• pass VM_RETENTION_PERIOD through the service environment and reference it inside the inline shell with quoting

Why

Monitoring retention needs to vary by plan while preserving the current defaults for existing Docker Compose deployments.

Closes #184 (closed) Related to #14

Validation

• bash tests/compliance_vectors/check_compose_retention_config.sh
• same retention config check inside docker:27.3 with bash, jq, and docker-cli-compose
• git diff --check
• glab ci lint .gitlab-ci.yml --repo postgres-ai/postgresai
• initial REV review: !259 (comment 3307455011)
• first rerun REV: !259 (comment 3307477586)
• second rerun REV: !259 (comment 3307491831)
• final REV rerun: !259 (comment 3307503694) (0 blocking)

Current Gates

• Pipeline: https://gitlab.com/postgres-ai/postgresai/-/pipelines/2494144755
• Reviewer requested: @Sarumyan9999
• Cap PG review-status thread still unresolved pending human review/approval