Compile bilbiography for analyses

[adelavega]

Closes #543

Looks like CSL is a good JSON standard for citations. In a .json file, I will keep a list of citations, with keys representing either tools, datasets or extractor_names. I don't think there should ever be clashes.
Then, an API endpoint will simply take a base set of tools + the dataset_name + extractor_names in an analysis, and return a CSL-JSON.

We could additionally try using citeproc-py to render a bibliography in APA or other format. However, the python library looks a bit clunky and not actively maintained, so I'm weary of using it. Also, this limits the formats we can provide (as it would require going back and forth, and doing it only on compilation).

Instead, I think its best if we use citations.js, a much more actively developed library, to render / compile the bibliography. This way, users could select a format, and compile it, including BibTex or another common format.

[adelavega]

To add more detail as to how bibliography.json is being populated. Zenodo natively exports to CSL-JSON, which is nice. For papers, I am importing to a local Zotero library, which can then be easily exported to CSL-JSON.

[codecov-io]

Codecov Report

Merging #547 into master will increase coverage by 0.52%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master     #547      +/-   ##
==========================================
+ Coverage   84.03%   84.56%   +0.52%     
==========================================
  Files          38       39       +1     
  Lines        1616     1671      +55     
==========================================
+ Hits         1358     1413      +55     
  Misses        258      258

Impacted Files	Coverage Δ
neuroscout/resources/analysis/__init__.py	100% <ø> (ø)	⬆️
neuroscout/resources/__init__.py	100% <ø> (ø)	⬆️
neuroscout/resources/analysis/schemas.py	97.29% <100%> (+0.19%)	⬆️
neuroscout/resources/analysis/bib.py	100% <100%> (ø)
neuroscout/resources/analysis/endpoints.py	90.57% <100%> (+1.32%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update fbc9e1a...c493401. Read the comment docs.

[adelavega]

This is almost done. @tyarkoni @rwblair what do you think?

Still need to add many more references (most glaringly for extractors). I decided to keep this info out of the db, and simply in a JSON file btw.

Unfortunately, I could not get citation.js to work, so instead the API returns APA formatted strings. That package seems to be in active development, so I'm hopeful in the near future I can instead pass CSL-JSON, which the frontend can then convert on the fly to BibText, and various Text formats.

[adelavega]

I modified bibliography.json so that the second key refers to the feature name for extractors. .* matches all. This is for extractors like the PredefinedDictionary which have features with different references each.

For datasets, and tools, they always have a .* defined, and that is what is always matched.

[tyarkoni]

Looks great!