Kubernetes: worker pods CPU / GPU / disk report change

[epicfaace]

We should change how available resources (CPU / GPU / disk available) are reported by worker pods. Currently, I think a worker pod will just report the amount of CPU / GPU / disk allocated by that pod. Instead, the worker pod should report the amount of CPU / GPU / disk available by its parent node -- this number will accordingly decrease as more job pods are allocated and run on the node.