Fixes race condition when epoll_ctl() gets called after the socket has
been closed, which may affect a different socket created by another
thread meanwhile.

When rpc_reconnect_requeue() gets called from inside nfs_service(),
the NfsInputStream can stall completely because the old socket has
been unregistered from epoll automatically, but the new one has never
been registered.  Therefore, nfs_service() will never be called again.

This kludge attempts to detect this condition by checking
nfs_which_events()==POLLOUT.

https://bugs.musicpd.org/view.php?id=4081

SocketMonitor::Cancel() does not actually unregister the socket; it
only disables the event.

When mounting had not yet finished, SocketMonitor::IsDefined() was
always false, due to the workaround at the beginning of the function
that calls SocketMonitor::Steal().  This commit drops the IsDefined()
check because it was never necessary and breaks reconnect.

nfs_destroy_context() will invoke all pending callbacks with
err==-EINTR.  In CancellableCallback::Callback(), this will invoke
NfsConnection::DeferClose(), which however is only designed to be
called from nfs_service().  In non-debug mode, this will leak memory
because nfs_close_async() is never called.

Workaround: before nfs_destroy_context(), invoke nfs_close_async() on
all pending file handles.

The method NfsConnection::CancellableCallback::Callback() will always
invoke NfsConnection::Close() on the file handle, even if the void
pointer is not a nfsfh.  This can happen if the Open() was not
successful, e.g. when the file does not exist.

During the NfsLease::OnNfsConnectionFailed() call, the old (defunct)
nfs_context may be used to close file handles.  Such code does not yet
exist, but will be added soon to fix other bugs.

Fixes potential second mount attempt after the old connection to the
NFS server was shut down.

If an async opertion is in progress, nfs_close_async() will make
libnfs crash because the RPC callback will dereference an object that
was freed by nfs_close_async().

Close the newly allocated file handle passed to the callback.

All locks are currenly held from only a single thread (the IOThread)
and thus we don't need the Mutex.

This allows eliminating the indirection code from the NfsConnection
destructor.

Reduce Remove() overhead because we don't have to walk the list to
find an iterator by reference.

Since BroadcastError() calls BroadcastMountError(), there's no need to
check mount_finished here.

While reconnecting to the NFS server, the socket is -1, because there
is no socket.  Fixes a potential crash (assertion failure).

More robust and cancellable.