The sharder aborts cleaving shards if the root DB does not have an
own_shard_range (see Related-Change). Previously, the cleaving was
aborted *after* a shard broker had been created, which potentially
leaves an unused shard DB in a handoff location. This shard DB would
eventually be replicated and cleaned up by the replicator, but its
creation can be avoided by checking the root DB own_shard_range
earlier.
Change-Id: I05ec542faa452db2053302910d34b93f0abd10a7
Related-Change: I68c3e7558d04e8dcd5874f80b4763ec0b820331d
6609fcd9a1