Usage of updater in VersionedUnpickler is undocumented and maybe redundant

[kitchoi]

The VersionedUnpickler accepts an updater object:

apptools/apptools/persistence/versioned_unpickler.py

Lines 111 to 113 in 632a4bf

	def __init__(self, file, updater=None):
	Unpickler.__init__(self, file)
	self.updater = updater

This was supposed to be its "interface":

apptools/apptools/persistence/updater.py

Lines 9 to 17 in 632a4bf

	class Updater:
	"""An abstract class to provide functionality common to the updaters."""
	def get_latest(self, module, name):
	"""The refactorings dictionary contains mappings between old and new
	module names. Since we only bump the version number one increment
	there is only one possible answer.
	"""

But from where it is used, there are more features expected of this interface, e.g. it is expected to have an attribute called setstates which is a dictionary:

apptools/apptools/persistence/versioned_unpickler.py

Line 157 in 632a4bf

fn = self.updater.setstates.get((module, name), False)

The structure of this dictionary is not documented (and there seem to be no tests for it). It seems to be used for monkeypatching a class __setstate__ (a global state that does not belong to apptools!), but it may not be restored.
Monkeypatching here in the code path if VersionedUnpickler.updater is not None:

apptools/apptools/persistence/versioned_unpickler.py

Lines 166 to 167 in 632a4bf

	# hook up our __setstate__ which updates self.__dict__
	setattr(klass, "__setstate__", __replacement_setstate__)

To be restored if some other unpickler without an updater came across the same class:

apptools/apptools/persistence/versioned_unpickler.py

Lines 145 to 148 in 632a4bf

	# restore the original __setstate__ if necessary
	fn = getattr(klass, "__setstate_original__", False)
	if fn:
	setattr(klass, "__setstate__", fn)

It is possible that no one uses an updater with the VersionedUnpickler so this monkeypatching is never exercised. I have not checked if there are other uses of an updater anywhere else. It could be a feature that can be removed.