RuozhiBench: Evaluating LLMs with Logical Fallacies and Misleading Premises