Improved the playlist processing by adding in MATH statements to scale the recipes to larger lists, and then introduced the concept of a local python indexed array of tracks. This gives the AI LLM a simple array [1,2,3] instead of the full Navidrome track IDs [bribkwebvwbvi,birubf8hg4eriobug,uefw9fbgr] etc, which it found hard to reliably sort without dropping a character here or there.