Asymptotic properties of Monte Carlo methods in elliptic PDE-constrained optimization under uncertainty