Add routing logic to UUID retrieval for sharding

[lkatalin]

Currently UUID parsing prepends a TreeID (of zero) to a UUID to create an EntryID, and returns that to the user. UUID lookup can still be performed as usual using this EntryID, because the TreeID is actually discarded during the query. We should instead make sure the treeID is correct when creating the entry (perhaps in #630 ), and use the TreeID to query the correct tree during UUID-based artifact retrieval.