Virtual Kubelet: Pod Volumes + Mounts

[MbolotSuse]

Currently, the Virtual Kubelet removes any volumes/volume mounts from a created pods configuration before syncing the pod to the host cluster. However, this is not acceptable outside of the POC - users should be able to run pods which use volumes.

Instead, for volumes that are derived from API resources (such as configmaps, secrets, service account tokens, API Server CA, projected volumes, pvcs) these resources should be synced from the virtual cluster to the host cluster. The version of these resources in the host cluster should have a name unique to the virtual cluster (so that the configmap "example" in namespace "default" can exist in two different virtual clusters in the same tenant). If users update the source resource, the new version should be automatically synchronized to the host cluster. If these resources are removed from the virtual cluster, they should be removed from the virtual cluster.

For volumes that are derived from resources outside of the cluster (e.x. nfs), these resources should be left as-is (allowing users to use properly named values from the host cluster).

[galal-hussein]

I think we can close this issue now, since we have synchronized pod mounts including configmaps/secrets and projected volumes as well as the service account tokens , lets follow up the pvs/pvcs in its own separate issue

#141