two proposals are presented:
-Proposal 1 corresponds to the case where a single general dataset is described, with declinations appearing at the distribution level.
-Proposal 2 corresponds to the case where several datasets are described, the distributions presenting mainly information on the format.
All comments are welcome!