Model-based Clustering using Automatic Differentiation: Confronting Misspecification and High-Dimensional Data