use WebSocket for Real-time reception

[czkoko]

Due to the influence of computer performance and model size, the waiting time may be a little long. Can we use WebSocket to communicate to get real-time line-by-line display code?

Although the total time of code generation will not change, it can improve the user experience. At the same time, the relevant code of vscode-fauxpilot also needs to be modified.

Here is an example of using WS for ggml model in c++.
https://github.com/monatis/llm-api