Inserts an empty scene after `afterSceneId`, or appends it when no id is given. The new scene inherits the preceding scene's `voiceDescription` so one character stays consistent, and defaults to a 6 second slot with empty prompts. Free: nothing is generated until you fill it in and regenerate its image. Adding a scene invalidates the stitched render, so `renderUrl` is cleared. A video holds at most 50 scenes.