Maximizing the Probability of Fixation in the Positional Voter Model