Prompt-Tuned LLM-Augmented DRL for Dynamic O-RAN Network Slicing