On Tackling Explanation Redundancy in Decision Trees