Avoid taking the peers write lock during event processing

Because the peers write lock "blocks the world", and happens after
each read event, always taking the write lock has pretty severe
impacts on parallelism. Instead, here, we only take the global
write lock if we have to disconnect a peer.