Eight Methods to Evaluate Robust Unlearning in LLMs