Fix sinfo -R to print each down/drained node once, rather than once per
partition. This was broken in the sinfo change to process each partition's
information in a separate pthread.