OBLIVIATE: Robust and Practical Machine Unlearning for Large Language Models