Images should not be simultaneously embargoed and not embargoed.

[matentzn]

We have one case of a dataset that contains both embargoed and non-embargoed images:

TrumanWood2018public contains two images that are also in TrumanWood2018. In P2, datasets that are embargoed are deleted in their entirety - which means those two images get deleted, when TrumanWood2018 gets deleted.

This pattern should be replaced by another approach - I suggest moving the non-embargoed neurons over to the public set rather than referencing them from both.

[dosumis]

This is a more general issue for data coming from CATMAID where the same neuron often gets referenced in multiple datasets. @Robbie1977 not sure of best solution here apart from adding dataset links to embargoed datasets only after setting to production.