Estimating individualized treatment rules in longitudinal studies with covariate-driven observation times