Manage per-model branches in a single chat

Will be a powerful workflow when combined with:

https://feedback.typingmind.com/p/be-able-to-cancel-a-primary-response

Each time we select a primary LLM from the parallel mode, a branch based on the replies of that LLM is created (e.g., the “GPT-5“ branch). Further, when we go back to the parallel mode (i.e., unselect the “GPT-5“ branch), we are in fact going back to the “parallel branch“.

Thus, each time we send a parallel query or select a primary branch, we create a node in the conversation tree, and this tree should be saved and be interactable in the UI.