Refactor Get started with a GH issue flow

As a developer, I would like to refactor the API flow for the "Get started with a GH issue flow" to get better scalability and performance.
3 backend calls are made for this functionality:
1. Frontend to Backend LLM to retrieve GH issue number
2. Frontend to Backend Github API to retrieve issue description
3. Frontend to Backend LLM again to retrieve solutions for the issue

At present, these calls are all initiated in the frontend sequentially. We would like to move this logic to the backend so that there is no back-and-forth communication between the frontend and the backend, and from the FE's point of view, there is just one initial backend call with the user query, and the response would be the solution of the issue.

No files or tests are named. Start by locating the frontend flow for “Get started with a GH issue” and the backend entry points it calls, then trace the three sequential operations: issue-number retrieval, GitHub description retrieval, and solution generation. Done means the frontend makes one backend request with the user query and receives the issue solution in the response.

Written by the indexing model from the issue text.