Rejo had code with these conditions:
| * ! ^FROM_DAEMON
| * ! ^Precedence: (bulk|junk)
The second of those two conditions is redundant, as anything that would fail
it (by including the regexp) will have already failed the first condition,
because the expansion of ^FROM_DAEMON already includes
^Precedence:.*(junk|bulk|list)
That of ^FROM_MAILER does not, so if you wanted to eliminate anything
matching ^FROM_MAILER, those two precedences, and nothing else that is
in ^FROM_DAEMON but not in ^FROM_MAILER, this would have made sense:
* ! ^FROM_MAILER
* ! ^Precedence:(.*\<)?(bulk|junk)
But if you've eliminated anything that matches ^FROM_DAEMON, items with bulk
or junk precedences are already gone.