Add clinical trials data to Crossmark custom_data

[gnott]

Re issue https://github.com/elifesciences/elife-crossref-feed/issues/146

CrossMark program, if the depositor is participating, can also accept clinical trial data in the <custom_meta> tag of the Crossref deposit.

The ClinicalTrial object will be populated when article XML is parsed.

The elifecrossref/clinical_trials.py module is new, which includes

• deciding whether to add clinical trials to the Crossref deposit
• parse the Crossref clinical trials registries XML file
• download the Crossref registries XML from them
• set the XML tags in the Crossref deposit to hold the clinical trials data

In order to download the registries XML from Crossref, requests library is now a requirement for this library.

The hook to add clinical trials is added to the elifecrossref/crossmark.py module.

There are tests to cover

• adding a clinical trial where the DOI needs to by looked up by name
• adding a clinical trial where the DOI was specified directly in the article XML file (when it is crossref-doi as its source_id_type)
• tests for parsing the registries XML file into a name to DOI map

[coveralls]

Coverage remained the same at 100.0% when pulling 606f40019dcfec6dd9915d6581e6661eaf01f4da on clinical-trials into 31ab5ba778e83fcaef50ec5ce3c44ce7330a1b63 on develop.

[gnott]

I will also mention, I didn't add an exception handling around getting the registries XML, so I guess we'd rely on the exception being caught upstream from this. A possible addition could be to check the status_code of the GET and log some things here.