An extension of $CVaR$ using non-precise a-priori distribution and its application (Decision making problems with uncertainty)