Restart stopped worker services before creating new ones

When a hosting platform makes service creation slow or failure-prone, a fleet controller can reduce scale-up latency by retaining stopped worker services and restarting them before cloning new ones. Reconcile each reused service to the current immutable image and configuration, require fresh readiness before routing, and retain enough namespace headroom for draining replacements and failed creations. This trades lower scale-up overhead for persistent service-definition sprawl and a larger reconciliation surface; it does not remove container startup time or make stale configuration safe.

Applies when: Autoscaling manages individually named service instances that can be stopped and restarted. Keywords: autoscaler stopped workers warm pool service reuse startup latency reconciliation