Seed reconciliation might break calico networking

[databus23]

We used to support switching from our kubenet/flannel based networking to something that supports network policies like callco.

We use this in the webshell clusters and there is a small number of customer who also went down that path.

With seed reconciliation there might be a conflict:

In order to switch to calico networking the flannel ds in kube-system needs to be removed and a calico-node daemonset needs to be deployed.
But the reconciliation will bring back the flannel daemonset and then we have two conflicting cni implmentations messing up the nodes cni config.

We need to come up with a way how we can continue to have modifications to the cni config that is deployed to nodes.