Connect flexible inference bench with CServe deloyment config

[xinli-sw]

Currently deployment config in cserve is laied out as such
https://github.com/CentML/cserve-product/blob/main/backend/cserve_backend/interface.py#L178

We should find a way to connect the output of flexible-inference-bench to PlanOutput