On sequential and nonsequential D‐optimal experimental design