ASSERT: Automated Safety Scenario Red Teaming for Evaluating the Robustness of Large Language Models