In Sylva, workload cluster values specifically indicate the version of k8s to use (`k8s_version_short`) and hence the "upgrade Kubernetes" step is not implicitly done automatically when doing an "upgrade Sylva" operation. Our CI should reflect that: * not upgrade workload clusters in pipelines that test sylva upgrades * rely on separate pipelines to test workload cluster kubernetes upgrades for k8s n-2 -> n-1, and n-1 -> n