Mediation analysis in a case‐control study when the mediator is a censored variable