Balancing bias and precision in capture‐recapture estimates of abundance