feat: Jan allows users to edit AI responses inline during conversations

[eckartal]

Problem
Users often find that the responses not entirely suited to their needs.

Users are able to edit the JSON file to alter an LLM's memory, but this process is not accessible from the UI. Users want a more direct and intuitive way to edit LLM responses inline.

Detailed user feedback: https://www.reddit.com/r/LocalLLaMA/comments/1cdk5nr/comment/l1q41wz/?context=3

Success Criteria
Implement an inline editing feature that:

• Allows direct modifications to AI responses within the UI.
• Ensures these edits adjust the AI's "memory" for future responses.
• Adapts the AI's response style based on user edits.

Additional context
A user comment: "just tested with editing the JSON, and it does change the 'memory' of the LLM as I hoped: https://imgur.com/BDtB5na"