feat(policy): update output handling for lint and eval

[Piskoo]

Update how eval error are split when returned from regal. Allow output flag in lint.

[jiparis]

Thanks!

[Piskoo]

I've changed the issue to include some changes that I made along the way, but in the end I didn't find a good way to refactor auto detection loop or any other code. Most tries ended up in making the code more complex and less understandable.