A few corpus-cleaning helper tools.

mass-find-nonspam

	Find non-spam, or otherwise suspect, messages in a spam corpus.  Has
	patterns which will match a good subset of typical newsletters found in
	trap data.

remove-tests-from-logs

	Guess.  Simply name the tests on the command line, reads log
	from stdin, writes rewritten log to stdout.

uniq-mailbox

	Trim duplicate mails from a mailbox.