[schema-manager] risk of running out of db memory with large numbers of vnodes

[bowrocker]

As @kellymclaughlin noted in MANTA-3896, the schema-manager should be refactored to process large numbers (> 10,000) of vnodes, otherwise we risk OOM errors on the database.

This should be able to be accomplished in the code, but splitting the list of vnodes returned from the fast query: here.