DITTO: A Spoofing Attack Framework on Watermarked LLMs via Knowledge Distillation