Minimax plans of regression experiments taking into account systematic errors