Since we're not using the message-ID for dupe checking or threading and we
already have a mechanism in place for body text dupe detection (via CRC-32)
the hash mechanism is providing an unnecessary burden on the mail database.