For CoT models, add the ability to hide their thinking process.

Gemini's 2.0 Flash Thinking Experimental model is designed to output their whole thinking process by default. This usually would be a collapsible menu, but for TypingMind, it just outputs everything alongside the thinking process making it quite annoying to read it's intended answer. I do not know if it's the same for O1.

A collapsible menu should be added, similar to how they would do it in Google AI Studio.