get_pmc_citeproc: switch to Literature Citation Exporter

[dhimmel]

NCBI Citaiton Exporter was taken offline without notice:
https://twitter.com/dhimmel/status/1061787168820092929

[dhimmel]

@agitter this should take precedence over other PRs, since currently any pmcid citations will fail.

Currently this PR only switches get_pmc_citeproc and not get_pubmed_citeproc to use the Literature Citation Exporter. We can migrate get_pubmed_citeproc in the future, but it's not pressing.

Once merged, we should update rootstock and release Manubot 0.2.1 (I will start preparing release notes).

[agitter]

I know we want to merge this quickly, so I would be okay merging as-is. However, the error message for an invalid identifier that begins with PMC and an empty or non-numeric suffix is not very informative.

$ manubot cite pmcid:PMC
## ERROR
'PMCID'
[]
## CRITICAL
Failure: exiting with code 1 due to logged errors

Other bad example cases include pmcid:PMCA and pmcid:PMC123A. The API provides a different response for invalid numeric identifiers, like pmcid:PMC3, which is caught and explained well by Manubot.

[dhimmel]

Ah interesting. When querying https://api.ncbi.nlm.nih.gov/lit/ctxp/v1/pmc/?format=csl&id= the following JSON is returned:

{
    "status": "error",
    "reason": {
        "id": [
            [
                "This field is required."
            ]
        ]
    }
}

However, when querying an invalid id such as https://api.ncbi.nlm.nih.gov/lit/ctxp/v1/pmc/?format=csl&id=3, the following XML is returned:

<pub-one-record-set xmlns:xlink="http://www.w3.org/1999/xlink" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:p1="http://pubmed.gov/pub-one" p1:error="PubOne failure: empty XML."/>

[dhimmel]

@agitter I made the error checking and messaging more robust. Take a look again?

[agitter]

Thanks, this version looks great. I tested with PMC0, PMC1, and PMCA to see the various error handling paths.