Prevent data citations from getting stripped in the final publication

[hmstepanek]

Problem
Data citations (DataCite + others) are stripped somewhere between the author's MS and Crossref (& other citation repos). Publishers run citation tools that check the author's citations before publishing, if the DOI can't be found, if all the data isn't present for the DIO that is needed to create the citation, if there is no DOI at all, etc then that citation may be stripped from the publication.

Solution
W/ a small change, each of the following 4+ groups can unilaterally reduce or eliminate this: Submission tools, Publishers, Cite-checking tool for publishers, Crossref, and Authors

1. Submission tools: Check for data cites in submission; sleep then check Crossref to see if it's there; ping authors+Crossref if it hasn't happened.
   [Todo: check script, cron script]
2. Publishers: Doublecheck any cites that your cite-checking tool doesn't like; if it's a datacite, handle properly; dblcheck the cite in Crossref isn't empty. Doublecheck that all cites in original MS end up in some form in Crossref.
   [Todo: Write doublecheck scripts, custom handler]
3. Cite-checking tool: Handle datacites (and software cites) properly. Talk to publishers facing this problem, include them in your smoke tests.
   [Todo: update these tools; both publishers and tool-makers can check to see if this is working]
4. Crossref: Doublecheck that data + software cites in original MS show up, aren't empty.
   [Todo: Write doublecheck scripts]
   A. for 2+3+4 : if missing and desired, generate DOI on submission
   B. Authors: After submission, sleep then search to see that the citation list associated w/ your article is complete / includes your datacites. Ping publisher? if this isn't working
   [Todo: make this search possible + easy; add "fix my datacite" API/email to Xref]
   C. Curious researchers, DataCite itself?: Write a bot that looks for all DataCite refs and check to see if they're properly listed in papers / associated w/ their citing papers in various repositories
   [Todo: write, run bot]
   D. Use Synapse to generate DOIs if needed-it has a very nice api for creating these.

[hmstepanek]

I think there's a role Hypothesis could play in this which is to notify users in the review process that the DOI cannot resolve and even that the DOI they used doesn't have the necessary information associated with it for the publisher to create the citation in the published draft. I think this would ultimately end up looking like a bot that runs on top of Hypothesis similar to Scibot that calls into another tool's api that does these checks and creates an annotation if it finds a problem.

[hmstepanek]

@meredithslota