Why did Davidski remove the Proto-Villanovan sample from the IA_Republic average even though it's not really an outlier? With it they cluster more firmly with Italians like it should.