Imitation Learning from Imperfection: Theoretical Justifications and Algorithms