Inference Extension: Extract model name from request body

[sjberman]

In order to support model redirects and traffic splitting, NGINX needs to be able to extract the model name from the client request body. It will then get used to set the X-Gateway-Model-Name header in the request to the Endpoint Picker (EPP) for it to make a decision on which endpoint to route the client request to.

This story is to simply write the module to extract the model name. A followup story will actually use this.

Acceptance Criteria:

• Create an NJS module that can extract the model name from the client request body
• The NJS module should be built in to the NGINX container images

Developer Notes:

• OpenAI uses a request body that should look like the following when a client makes a request to an AI workload:

{
  "model": "gpt-4o",
  "messages": [
    {
      "role": "user",
      "content": "Some question..."
    }
  ]
}

Design doc: https://github.com/nginx/nginx-gateway-fabric/blob/main/docs/proposals/gateway-inference-extension.md