Optimal timing of carbon capture policies under learning-by-doing