State:
This is a first cut at storing and retrieving data. There's no doubt it needs more work, 
and the data is very incomplete, but it does do simple things (see EXAMPLE) correctly.

Some things that don't suck
 * identifying the build that is making the query (for the purposes of knowing if a partial
   is valid)
 * being able to expand the json for a release and covert it into snippets and xml

Some things suck or maybe suck
 * lots of data manipulation, need some Classes and sqlalchemy goo to handle that all
   automatically; similarly, need to shove some code out to 
 * same data is stored lots of different ways in different places, cull the middlemen if 
   they aren't needed
 * do we still like json encoded into text field ? does it create more hassles than it solves ?
 * the db are just flat tables, s'ok ?
 * no support of Rob Strong's new attributes