more robust detect_separator

[AlexisEidelman]

see de1e729

[ColinMaudry]

(@noirbizarre informed me of your experiments)

I work on a similar project, with the same data (except I work from the RDF version of the metadata).

A possibility is to check which of the typical delimiters is present as many times in the first 4 rows.

[Leobouloc]

@ColinMaudry Thank you for your advice. This should be pretty straightforward to implement. I'll try that when I get a chance. Feel free to make pull requests in the future (hopefully the code will be cleaner and more understandable).

[ColinMaudry]

I unfortunately don't speak Python, I'm doing it in bash, using sed, csvkit, and csvtool.