robj / TODO


* Figure out how to let someone specify arguments in the uri and properly
  cache instances based on any arguments.
* Possibly mark invalid instances as such. When an resource is deleted or
  refreshed there may still be references to the previous version of this
  resource. They could could be marked as invalid, letting the user know that
  they are out of date if the user attempts to access any of the attributes.
* Allow consumers of rObj to provide a type map or document class for each URI.
* Figure out how to make running dir on an instance not fetch all attributes.
* Provide a way out when rObj doesn't do the right thing.
* Add support for Cookie based authentication, in particular OAuth.
* Add support for json data in addition to xml.
* Add download callbacks (maybe just for file downloads).
* Add redirect and possibly error handling callbacks.
* Maybe switch to deffered fetching of resources. Particular use case is wanting
  to delete all items from a collection one at a time. This doesn't require
  fetching each instance, just calling do_DELETE on the instance's uri.