sendout.c:_message_id(): tweak [f2a5e4e]..

sendout.c:_message_id(): tweak [f2a5e4e]..

Well i missed that we test the value that i had adjusted
further on, so revisit that.
(I really have to leave this codebase alone a bit now.)

Note there is a dangling bug that i don't yet understand:
sometimes we generate

  Message-ID: <20150514173456.Ro3Tx7HAhdbQuyPVKJ7b@yandex.com>

i.e., without user, even though it should be

  Message-ID: <20150514174207.7wU1BGYxBgjS%sdaoden@yandex.com>

i.e., the __sendout_ident hack should work as it always does.
I have seen this multiple times in the last circa two weeks, but
whenever i had immediately set *debug* and tried again, it cannot
be reproduced at all!  Maybe a compiler error?  No idea yet!
(Besides it would possibly be better to simply drop
__sendout_ident again and only print TIME.RANDOM@HOSTNAME as most
other MUAs do, we deliver something in From: anyway, do we?
I think that's where we'll end up… when that myorigin() desaster
is finally gone.)