Log fetching and analysis

[thanethomson]

It's critical to the debugging process to be able to analyze nodes' logs, so we should look at:

• being able to fetch logs from nodes (this could be a simple Ansible script to just pull all the Tendermint log files from all nodes), and
• being able to analyze them effectively.

A simple, quick 'n dirty approach to enabling log analysis may be to just output all Tendermint logs as JSON and import them into a PostgreSQL database once we've fetched them. I haven't used Postgres' JSON functionality extensively, but I'm somewhat confident it'll allow for quicker log interrogation and debugging than greping through by hand - especially if we can create indexes on JSON fields.