cosim: Fix race-condition in cosim

[Giftzwerg02]

This PR should (finally) fix the flaky Ppc64Cosim tests which rarely appeared.

I found a race-condition, more specifically a lost wake-up, exactly as described here. I forgot to re-aquire or keep the mutex lock on the broker-side when sending a signal (or rather, when changing the dependent variable) to the client that it can write to the ring-buffer. In rare cases, the client checked if it can write (and saw it couldn't), then the broker sent a signal (which wasn't yet caught) and only then did the client start the conditional wait.

Simply keeping the lock while changing the counter fixes this problem, i.e.:

pub fn end_read(&mut self) -> Result<()> {
    self.read_idx += 1;
    self.count.fetch_sub(1, Ordering::SeqCst);
    self.notifierpost()?;
    self.notifier.unlock()?; // <- mutex lock is now kept until here
    Ok(())
}

This PR also uses "robust" mutex locks to better check whether a QEMU client has crashed:

PTHREAD_MUTEX_ROBUST
If a mutex is initialized with the PTHREAD_MUTEX_ROBUST attribute and its owner
dies without unlocking it, any future attempts to call pthread_mutex_lock(3) on
this mutex will succeed and return EOWNERDEAD to indicate that the original
owner no longer exists and the mutex is in an inconsistent state. Usually after
EOWNERDEAD is returned, the next owner should call pthread_mutex_consistent(3)
on the acquired mutex to make it consistent again before using it any further.