Is your feature request related to a problem? Please describe.
Currently there is no way of tracking requests for llama-cpp-server afaik (I may miss a related feature). It would be really good if we can put custom IDs/tokens for client side that will be returned by the server. That would be very useful for backtracking, handling async requests, etc.

Describe the solution you'd like
Usually Request-ID or Correlation-ID. A related read about using which one. This is easy to implement, just returning an id in the response headers; however, it would significantly impact the users imho.