Since we started to deploy kube-system, flannel and wormhole as daemonsets we forgot to handle the upgrade case so when upgrading from 1.24.X -> 1.24.y or 1.25.z the daemonsets stay as is and are not upgraded.
I guess we finally need to merge #679
If this is not feasible we need to think about a stopgap solution and handle at least updating images during cluster upgrade for kube-proxy.