Refactoring:
* New `Chapter' class extracted to separate HTML parsing from
adapter output preparation and story-wide metadata collection.
* Lazy-parsing and fragments caching is done at Chapter level.
Improvements:
* Rating "adultness" and label-to-title mapping made configurable.
* Fix chapter number detection when title contains large numbers
(ex.: http://www.masseffect2.in/publ/19-1-0-2934).
* Add mechanism for detecting series-like stories with no chapter
numbering and extracting correct titles.
* Fix number format for generated chapter titles.