Robustness at different parts of the codemeta.json creation

[maelle]

codemetar gets data from GitHub API, soon from a csv with information about rOpenSci reviews cf #23 .

It'd make sense to check for the presence of an internet connection, to try a few times, and to output NULLs instead of errors.

[maelle]

Also be more robust to hitting GitHub API rate limits, especially in the absence of a token.

[maelle]

• if rOpenSci review badge, rOpenSci review info gotten from a JSON on the web.
• if GitHub repo, Preferred README, and repo topics, gotten via GitHub API (gh::gh()) . In a try().

[maelle]

@sckott have you actually observed many failures when building the registry?

[sckott]

not many no. There was that PR #247 and #236 , but those fixed now.

The biggest thing I can think of is the keywords problem, which I don't really have any ideas how to fix since its so intermittent

[maelle]

fixed iv dev by #255