Efficient and robust approaches for analysis of sequential multiple assignment randomized trials: Illustration using the ADAPT‐R trial