Build configuration for whether to run lint and whether errors stop builds

[twaugh]

e.g.
"lint": "yes",
"lint_fail_on": "errors" / "warnings" / "none"

Would also be useful to be able to configure which set of rules to use but there aren't standard ones yet as far as I can tell.

[TomasTomecek]

I would prefer having it in separate task to be honest. Since simple build task is quite stable (knock knock) it would be great if we weren't touch it much. For me, this means, that we'll have one stable build call and the other, improved call: lint & build, for testing.

The main reason why I'm being such a PITA is that I was so surprised yesterday, when it all broke down and I had no idea what happened.

Therefore I would rather go for incremental approch: add to existing (which is somehow stable and working), rather than alter existing (and break it possibly).

[twaugh]

One approach then would be to run the lint task in parallel rather than as the first link of a chain.

group([linter_task,
       build_image_task | submit_results_task])

This would mean the linter output wouldn't be able to prevent a build if there were errors, at least as things stand. Potentially we could abort the build_image_task if we made it abortable.