Clarifying Selection Bias in Cluster Randomized Trials: Estimands and Estimation