Show Open AI library can talk to Xef server.

[raulraja]

The current frontend code uses a custom request to the xef server.
This PR replaces that custom code for the TS openai library.

Currently failing with:

Access to fetch at 'http://localhost:8081/chat/completions' from origin 'http://localhost:8080' has been blocked by CORS policy: Response to preflight request doesn't pass access control check: No 'Access-Control-Allow-Origin' header is present on the requested resource. If an opaque response serves your needs, set the request's mode to 'no-cors' to fetch the resource with CORS disabled.

The goal here is to show a client with the openai library can target streaming with the Xef server.

[javipacheco]

🚀 LGTM!

[raulraja]

I guess we'll merge this after fixing the CORS biz and open a new Pr just for the streaming portion