consider changing the ids from auto-increment to full checksums to make distributed storage easier to organize

[CodyKochmann]

Distributed storage has been a long time todo for this db but the auto-incrementing functionality used for the keys made it hard to figure out what the right way was to evenly store things across multiple servers due to the nature of high demand inserts.

Switching to a full hash-based layout would work well both for the distributed setup and would bring the dbm based version of graphdb a bit closer in terms of engine design.