Text this: Pitfalls in interpreting calibration in comparative evaluations of risk models for precision lung cancer screening