Gravar-mail: Efficient estimation for the proportional hazards model with bivariate current status data