Add a "Split Segment" tool to break up long dialogue blocks & fix missed speaker changes

* The Problem: AI transcriptions often group long stretches of audio into massive, unreadable blocks of text. Furthermore, the AI sometimes misses a quick speaker change (like an interruption or fast back-and-forth) and lumps two different speakers into a single paragraph. Right now, fixing this formatting inside the platform is difficult or impossible. * The Fix: Allow users to click anywhere inside a text segment and hit "Split" (or use a keyboard shortcut like Enter) to instantly break that paragraph into two separate blocks. The system would automatically calculate the new timestamp for the split. * Bonus Benefit: This gives users total control over their transcripts. It makes fixing speaker labels incredibly fast and allows video/audio editors to create perfectly timed, bite-sized captions without having to export the text and fix it in a separate app.