Testing wrong return code when sd_journal_next() was not called as it
was not needed. When step_next == FALSE the EOF flag must be never set
since the journal EOF case is tested at the beginning of the function
and we're not advancing next thus staying on a valid cursor.

--
found by failing tests, should check them more often :-o
BTW I'm not adding any tests regarding the journal EOF case as they would be racy from nature on a live (system) journal.