fix: probe kube-state-metrics on its main port

[RomeoCavazza]

Summary

• send the /livez liveness probe to the kube-state-metrics exposition port (8080)
• retain /readyz on the telemetry port (8081) as documented upstream

Production evidence

• server-side dry-run succeeded
• targeted deployment rollout succeeded
• the prior pod had been repeatedly killed because /livez returned 404 on port 8081

Reference

https://github.com/kubernetes/kube-state-metrics#healthcheck-endpoints