Add scripts for evaluation with Microsoft Azure trace

[Flechman]

• Create a python script to generate requests with given request rate. The script writes the requests to a json file, which will be retrieved by a C++ request producer.
• Create a C++ request producer, which will simulate request arrivals given the json file with the requests. The same file contains the consumer, which waits for requests to be processed.
• The request timestamps come from the Microsoft Azure LLM inference trace, and are fused with prompt data coming from the ShareGPT dataset. Overall, the timestamps span an hour.

---

This change is