Add scheduled runtime DB retention/compaction

[techmore]

Problem\nRuntime logs and customer scan history are only pruned via a manual maintenance endpoint. For long‑running deployments, the SQLite runtime DB can grow unbounded, impacting performance and storage.\n\n## Evidence\n- runtime_db.apply_retention_policies exists but is only wired to /api/runtime/maintenance/retention (nmapui/handlers/routes.py).\n- No scheduled or startup invocation of retention policies.\n\n## Proposed Fix\n- Add a periodic maintenance task (e.g., daily or weekly) to run apply_retention_policies.\n- Expose retention limits in settings with sane defaults.\n- Log retention runs to runtime_logs for observability.\n\n## Notes\nThis is especially important for always‑on deployments scanning daily.

[techmore]

Tracking under #163 (Runtime data lifecycle).

[techmore]

Fixed in b02952d — added _run_scheduled_retention helper called from auto_scan_loop on every minute tick. It checks maintenance_retention_status snapshot and skips if <24 hours have elapsed. When due, it calls apply_retention_policies(compact=True), persists the result as a runtime snapshot, and appends a maintenance log entry. runtime_store is now threaded through build_runtime_bindings → start_auto_scan_thread → auto_scan_loop.