I see no reason why waitid() can't support other linux-specific flags
allowed in sys_wait4().
In particular this change can help if we reconsider the previous change
("wait/ptrace: assume __WALL if the child is traced") which adds the
"automagical" __WALL for debugger.