Chain of Natural Language Inference for Reducing Large Language Model Ungrounded Hallucinations