add provenance info to HAPI info response

[jvandegriff]

HAPI needs the ability to report where the data came from. Ideally this would be a DOI, and those are universal enough that we could add a field for that in the info response.

This means we should maybe revisit the ability for the server to return just the data with no header. Because it’s the header for a specific data request that should contain the DOI, or maybe a list of DOIs. Currently, most people don't really request the info for a specific time range - they just get it once to se what the parameters are. And some datasets use one DOI per file, or group of files. This one-DOI-per-file is not what is currently recommended by IHDEA, but actually there is not a solid agreement on that yet.

Maybe having a provenance block, and offer a list of suggested keywords:

• DOI (can be list)
• version (also a list?)
• modification date
• dependencies (can of worms)

[dlindhol]

We should consider looking at PROV-O for inspiration (e.g. names of properties).

[rweigel]

@jvandegriff keep in mind that CSV and Binary are transport formats, not file formats.

We should research the previous discussion on headerless as default before re-hashing. This has come up over and over.

A change in default would break many things.

[jvandegriff]

This needs longer discussion - possible other endpoint, or when getting data, add an include=provenance

Note that info responses can already include a resourceURL and also a citation

[jbfaden]

See also #217

[rweigel]

Conclusion #224 (comment)

For HAPI 4.0, if demand, consider adding provenanceSchema and allowing provenance to be an object.