* Allow different fields to be written out to the metadata log and the
statcache file.

* Implement a scheme for cleaning metadata log segments when metadata is
re-used.

* Improve the interface for remote upload scripts.  The current
interface is preliminary and is subject to change.

* Continue to investigate schemes for adding parity blocks for error
recovery.

* Rename cumulus to cumulus-backup, and write a new Python program,
simply called cumulus, to replace the functions in cumulus-util and act
as a wrapper for cumulus-backup.  Integrate automatic segment cleaning
and backup expiration.

* Write a module that allows duplicity storage backends to be used for
cumulus.