make easier to detect and remove bad keys from archive

[mmckerns]

Bad keys can come in the form of unhashable types... that can successfully be added to the underlying archive, but then certain methods fail when trying to use the dict interface to the archive. Failure can be that archive.__asdict__() throws an Error or appears empty.

See examples in #56, #64, #66.

[mmckerns]

Bad keys can be often removed with del archive[bad_key] or one of the 'private' methods like archive._rmdir(bad_key).

Keys that can be added to the archive, but are (for example) unhashable and thus can't be used in a dict view of the archive... should either not be allowed, or possibly more favorably, the 'dict' view of the archive should allow unhashable types.

Keys that are bad (but in the archive) should have a method that enables them to be detected (like archive.unhashable_keys() or archive.bad_keys() or something similar.

We should also be able to identify bad entries with archive.bad_entries() or something similar.

More care needs to be taken to not allow bad keys to be added, or if bad keys/values/items are in the archive, then it should be easy to detect and purge them (i.e. archive.clean()).

[mmckerns]

For example, the key [1,2] can be accepted as a key by many of the archives, but it throws a TypeError when the archive is viewed as a dict (as the key is unhashable).

Another case might be a corrupt file/entry.... these cases should get mocked up in test code.

[mmckerns]

pushing this just into the next release, but on the short list.