Enables text-to-image generation via RunningHub's seedream-v5-lite model, with support for submitting tasks and polling for results.
