A comparative study of alternative extreme‐value volatility estimators