Double robust estimation for multiple unordered treatments and clustered observations: Evaluating drug‐eluting coronary artery stents