multiple database support

[cyberhck]

There'll probably be a situation where we'll want to separate our read queries and write queries.

We need to have multiple database context, idea is to have 1 master db which only takes in writes, and N slave dbs which will be weighted and will be able to use weighted round robin for this.

Also add in a way so it's possible to do this, but if people don't want, should be easy to switch to simple 1 db as well.