to do for a first real version:
- link integrity vs iterate
- create method to re-create integrity refs for the whole site and
  integrate that into migration and the control panel
- what about ftp/dav?

nice things to have later on:
- use annotations in info adapter (info.py)
- update implementation notes
- it would be possible to cancel single removals while going through with
  the others, but only if a potentially uncaught (by other code, i.e. third
  party products) exception was raised on cancel;  atm it's unclear if this
  would be desirable, though...
- reintroduce objects names/path in portal message on cancel
- add warnings about moved images/pages
- factor out all archetypes related stuff using adapters etc (i.e. references)
- possibly replace atreferences by something more generic (five.intid should do)
- use pullparser.py instead of the newly written link-parser?

Things done and/or discovered in PLIP 13260 (winebeersprint 2014, saily, do3cc)
- restructured whole package
- using at and dx folders to allow setup of different specific tests in each
  folder.
- converted all doctests into p.a.testing tests, which run now quite fast.
- linkintegrity is a huge hack. we should think about a more cleaner and
  generic approach to implement such checks.
- currently we have one failing test on dexerity, which targets renaming
  documents. it has a decorator allowfailure to not break whole setup, we might
  need to fix that later.
