Comprehensive Evaluation of Machine Learning Countermeasures for Detecting Microarchitectural Side-Channel Attacks