We dont offer an aletartive runtime in the managed Kubernetes offering yetand, this is not planned in the short term, though we are looking ar katacontainer for other offerings on the longer term.
It is important to note that each kubernetes woker nodes are running on VMs in your project, therefore offering grat isolation between two cluster.
I am interested about more details on your use case ? Do you run a PaaS hosting different cutomers's code that require hard isolution within the cluster ?
Note that we are also working on offering smaller worker nodes,(still with guaranteed ressources) in a few months, would that be a valid workaround for your use case (playing with taints/label for affinity of pods for a given team/customer ?