GitLab

it was just too confusing on which one an editor is using, double comments
in jenkins, etc.

this is accomplished by several things:
 * required cops are just marked as severe, instead of using a separate
   config for them, and failing if anything shows up from that config
 * get rid of all the logic to only include certain directories for
   certain cops. turns out it's not _that_ ominous to correct errors
   across the entire repository before marking a cop as required.
 * but still auto-generate config to turn _off_ autocorrect for
   non-severe cops. this is important because auto-correct must run
   for entire files, and we don't want it auto-correcting optional
   things that you didn't touch.
 * update gergich to get more details from the parsed comments.
   this plus the prior point means we _don't_ have to have heavy mode when
   in autocorrecting, but we still display out-of-context lines that were
   autocorrected

this also makes it so we can use per-dir .rubocop.yml files again, so
take some of the exceptions out of the root and put them in their own
directory

Change-Id: Ie936d1a9920b68910acd250ba817c7b4a670b958
Reviewed-on: https://gerrit.instructure.com/c/canvas-lms/+/274394

Tested-by: Service Cloud Jenkins <svc.cloudjenkins@instructure.com>
Reviewed-by: Simon Williams <simon@instructure.com>
QA-Review: Cody Cutrer <cody@instructure.com>
Product-Review: Cody Cutrer <cody@instructure.com>