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.