Re-factor database creation code to support explicit version in schemas

[agstephens]

Proposed improvement for Ansible playbook to create/update the database:

1. All tables live under their own schema (so nothing at the top-level). Schema names:

   • v1_lite
   • v1_full
   • v2_lite
   • v2_full

2. The playbook should have a variable to define a list of versions, and then will create the schemas and tables for each.

3. Playbook should create, for both full and lite:

   • all partitions
   • all triggers
   • all indexes
   • all postgtres users (based on playbook variables)

4. Playbook should include tasks to do repeatable operations such as:

   • sorting (i.e. clustering) of each table (partition)