We should specify implications of identity comparisons on metadata #64
Description
We should describe in prose of the core spec the expectations for the following two otucomes of N metadatasets describing the same dataset:
- all such metadata instances share the same identifier
- not all metadata instances share the same identifier
We will want to avoid distinction without a difference, but also block re-use of identifiers where it is confusing or wrong.