More robust parsing of responses

The parsing functions (intchron_extract() and intchron_tabulate()) currently just extract specific elements from the listified JSON response based on what other functions needed. Specifically, they assume the important data is contained in a "records" array. But according to the schema (http://intchron.org/schema) and in implementing #5 and #6 I encountered other data structures.

Rather than continuing to pile kludges into intchron_extract(), we should have a generalised R representation of the response data. Probably with a dedicated S3 class as suggested by https://cran.r-project.org/web/packages/httr/vignettes/api-packages.html.

Start by reading intchron_extract() and intchron_tabulate(), then compare their assumptions with the response structures described by the IntChron schema. Review the linked httr API-packages guidance and the context from issues #5 and #6. Done means the response has a generalized R representation that supports structures beyond the records array.

Written by the indexing model from the issue text.