Improve code completion quality

To upvote this issue, give it a thumbs up. See this list for the most upvoted issues.

After #12 , ECA code completion now allows the use of any supported/custom model and prompt, but its defaults are not as good as GitHub Copilot or other models.

• model: This is probably related to the fact that most models are not focused on completion but on chat, I'm not aware of non ollama completion models open from anthropic/openai or github, that would be a good start.
• prompt: We use this system prompt which probably can be tunned for better results with specific models, would be nice to find some way to improve genericly like eca chat prompt.

Any help on this is welcome, let's discuss here.