outfactor source2dicts to core module

[lukavdplas]

Reader subclasses all implement more or less the same routine to extract documents from a source; this update outfactors that behaviour to the Reader class.

Specifically, it used to be that each reader subclass like CSVReader needed to implement source2dicts from scratch. This method now has an implementation on Reader which calls on some more specific methods like data_from_file and iterate_data (which are implemented by the CSV reader).

This makes the implementation of CSVReader, XMLReader, etc. a lot shorter. It also eliminates some cases where all these classes were implementing slight variations of the same behaviour (see below).

Changes:

• Better documentation on what is required to implement a Reader subclass.
• All readers now support these:
  • Including and omitting metadata dictionaries
  • required property for fields. (This was only supported in the XML reader.)
  • Order extractor