Feature: Support preempt/reclaim in network topology aware scheduling

[JesseStutler]

What is the problem you're trying to solve

Currently, when in preemption/reclaimation of network topology aware scheduling, we did not consider which HyperNode the Node belongs to. After preemption/reclaimation, the Pods in the Job will likely be distributed in different HyperNodes.

Describe the solution you'd like

We need to consider the network topology distribution of Node. Need to add implementation in

• [ ] preempt
• [ ] reclaim

Additional context

If there are other preemption/reclaim scenarios that have not been considered, please comment below

[JesseStutler]

/assign Would like to work on this