improve documentation/tests for corpus readers

[mssammon]

documentation is very incomplete for Ontonotes corpus reader -- no explanation of format or behavior wrt character offsets in original documents; no tests

[mssammon]

@sroy9 any questions about this?

[sroy9]

@mssammon working on it, no questions yet.

[sroy9]

@mssammon The Ontonotes reader seems to not work, tried a lot to get it to work. I have added a prospective test in a new branch 'sroy9_issuefix'. Test document added to resources/ontonotes. Has this reader been known to work for some document ?

[mssammon]

@mayhew @penghaoruo @danyaljj do you know of anyone using Ontonotes reader?

[danyaljj]

Not sure. @christos-c any ideas?

[christos-c]

I have seen it used for some experimental projects (like some version of the Temporal classifier) but nothing public (certainly not on the main Github repository anyway).

[mssammon]

@qiangning @666666fzl have you used OntonotesReader for temporal work?

[qiangning]

@mssammon I think no since I had never seen OntonotesReader until I looked it up just now.

[mssammon]

@penghaoruo have you used OntonotesReader for coreference work?

[penghaoruo]

@mssammon No, it is a CoNLL reader, not ontonotes directly.