HateBench: Benchmarking Hate Speech Detectors on LLM-Generated Content and Hate Campaigns