Super frustrated. I have found out that certain models are congesting the thread. now I need to rework the way I receive the tokens, the characters from llms, how I render them, and how I use highlighter so it doesn't re-render on every change the DOM and congest the thread #t3clone