Publication-level relationships between datasets

[afrubin]

Right now, MaveDB has publication information and a deprecation system for replacing datasets with errors, but I think we need an extra relationship between datasets that relates to the publication status of the experiment.

Specifically, this would allow us to track when the same dataset has been:

• Uploaded pre-publication
• Reported with a preprint
• Reported with a peer-reviewed publication
• Corrected post-publication (via official errata or otherwise)

I think that relying on the deprecation system might be too heavy handed, and it also lacks important contextual information about why a dataset is replaced by the other.

[afrubin]

To add to this, we may not want to actually "replace" the datasets as we do in the deprecation case (where the deprecated dataset is inaccessible except by linking to the exact URN), rather the intention is to be able to direct users to the best/most up to date version of the data. This will also have utility when figuring out what variants to expose in the clinical interface.