Fix: Make the library more robust for publishing

[satra]

1. updates datacite constructor to ensure

• contributorType
• support validation against jsonschema from datacite (this may not be authoritative and should be used with care)

2. ensure that round tripping of schemaKey creates the same object or raises a validation error

closes dandi/dandi-archive#347

[codecov]

Codecov Report

Merging #41 (d7b9a92) into master (054544f) will increase coverage by 1.20%.
The diff coverage is 97.39%.

@@            Coverage Diff             @@
##           master      #41      +/-   ##
==========================================
+ Coverage   94.51%   95.72%   +1.20%     
==========================================
  Files          11       11              
  Lines         857     1005     +148     
==========================================
+ Hits          810      962     +152     
+ Misses         47       43       -4     

Flag	Coverage Δ
unittests	95.72% <97.39%> (+1.20%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
dandischema/metadata.py	96.55% <95.23%> (+0.47%)	⬆️
dandischema/models.py	93.71% <96.42%> (+1.86%)	⬆️
dandischema/consts.py	100.00% <100.00%> (ø)
dandischema/datacite.py	94.79% <100.00%> (+0.96%)	⬆️
dandischema/tests/test_datacite.py	100.00% <100.00%> (ø)
dandischema/tests/test_metadata.py	100.00% <100.00%> (ø)
dandischema/tests/test_models.py	95.12% <100.00%> (+0.47%)	⬆️
dandischema/tests/test_utils.py	100.00% <100.00%> (ø)
dandischema/utils.py	100.00% <100.00%> (ø)
... and 1 more

---

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 caddcd3...d7b9a92. Read the comment docs.

[yarikoptic]

some minor comments from cursory review

[yarikoptic]

I know that it is just a code move.

But this better be cached or (even without any analysis of how often this is called) we might end up with all kinds fails upon some intermittent connection issues etc.
Or might be better to just include that file right here in the dandischema sources/distribution -- who knows what RFing they would decide to do in datacite etc. Then this file would always be with us and reliably available -- would make it more robust.

[satra]

that's why i have left that by default as False. we should have some trust no?

people can retry on http errors for anyone using the code. i will at least make it a permalink.

[yarikoptic]

so it could make it BarePublished<val>? shouldn't Published be added later so it would be PublishedBare<val>?

[satra]

that should not happen here. the second if is an equality check, in which case the only value tempval should have is Asset, thus turning it into BareAsset. anything else it should throw an error in the next few lines.

[satra]

@yarikoptic - feel free to merge this tomorrow if you have no other changes to make.

[yarikoptic]

If we talk about "more robust": I still think we should cache online fetching, and if someone is to retry -- it is for a fetcher . But that could be done later, so I am ok to proceed