Bubble up more information about Pod status in Model/Notebook/Server APIs

[nstogner]

A common case will be that a Pod requires a GPU but the scheduler/autoscaler is unable to place the Pod on a Node that has one. We shouldnt require users to go digging through Pod statuses and events to find this information. I think it should be shown with kubectl get models (and get notebooks, get modelservers) under the CONDITION column... see for reference:

k get models
NAME                READY   CONDITION
facebook-opt-125m   True    BuiltAndPushed
my-model            True    BuiltAndPushed

[nstogner]

Information about whether something is pending could probably come from the Job API. Perhaps a kubectl plugin would be better suited for further drill down… ie events and pod statuses (in order to avoid overactive reconcile loops and API updates).