Median procedural performance rubric scores improved from 16 to 26.5 after simulation-based mastery learning training (p < 0.001)