Quantile treatment effects and bootstrap inference under covariate‐adaptive randomization