server: Avoid accessing free'd thread pointers.

[Mike McCormack]

Eric Pouech wrote:
> IIRC, the issue in this code is that you access in _SAFE macro the next 
> field in the current cursor *after* the current cursor has been freed
> the issue is not that the next item has been freed while itering on the 
> current cursor
> (this was at least the issue I had)

It looks like kill_thread can recurse  if another thread is waiting on 
the current thread we're killing.

wake_up -> wake_thread -> send_thread_wakeup -> kill_thread

If the waiting thread is in the current process, and it's later in the 
list, I'm not sure anything stops it from being free'd.

Mike