estimate function for zero bootstraps

When we do the leave-one-out and related estimations, we don't want any bootstrap inference. Maybe we should alter the estimate function so that if b == 0, then bootse defaults to NA? Will any downstream functions need to be changed (e.g., print)?
