Add an example of serving multiple LoRAs per base model when serving multiple base models with BBR. Related to [this issue in llm-d](https://github.com/llm-d/llm-d-inference-scheduler/issues/415)