Comments on: Practical difficulties in testing identifiability of linear structural models.