A reconsideration of three-item analysis, the use of implied weights in cladistics, and a practical application in Gentianaceae