Enable KubeFlow During Cluster Creation

[VestigeJ]

KubeFlow should be a simple configuration add-on from the operators perspective to get up and running with a full RKE2 based machine learning environment. I'd like to propose that we do the hard work of roping in the KubeFlow manifests for auto-deployment similar to how we pass in args for the kubelet-args and kube-apiserver-args. I think this could couple very nicely with the same efforts of getting kubernetes GPU hardware enablement running with less confusion from our customers.

We would absolutely end up using this internally as well.