On the use of best tests to obtain best one-sided [beta]-content tolerances intervals--discrete case