[doberman] Make CrudeAnalysis, Refinery, etc generic enough to allow it to be used on deployments outside of OIL

rharper has been asked about the possibility of using crude/doberman on deployments outside of OIL for failure analysis.

This other team deploys openstack clouds via a bundle -> so, post processing of juju status, juju debug-logs, and juju-deployer console output

They may wish to use their own database and do not have pipeline stages (mostly just deploy)

----------------

This has taken on a higher priority now that there is talk of an analytics/big data lab. Also, I was approached at the CDO sprint by a couple of other groups (gnuoy/beisner) looking to use it in their openstack clouds.

It needs therefore would need to be sufficiently generic to be able to work without test-catalog if necessary. It would also be sensible to make it so it can use an alternative for jenkins an even lauchpad. (Interfaces needed here).

Also, we need tests, docstrings, documentation, etc

Finally, it should be able to jsonify the artifacts via another interface to logstash or whatever...

Crude, refinery, filing_station and oil-stats are being refactored to use a common object. Report-generator will also be altered to rely on these common components eventually.

Test-catalog and jenkins will also be written to function as plug-ins or backends, or whatever you want to call them, so an API will be needed.

Also, something similar for the database (could use either db or yaml... or both)

John Agosta mentioned some scripts he had written for filing a bug on launchpad which might be useful:

https://code.launchpad.net/~jagosta