Design of Experiments for Stochastic Contextual Linear Bandits