Good idea. In fact, even if bot_update did rebasing, I think CQ still shouldn't run when there is a merge conflict. Leaving Untriaged for CQ meeting, so I get more opinions.

CL: https://chromereviews.googleplex.com/489017014

The following revision refers to this bug:
  https://chrome-internal.googlesource.com/infra/infra_internal.git/+/c2a8a588f72bd598a0bd2660aff7d6074954d1be

commit c2a8a588f72bd598a0bd2660aff7d6074954d1be
Author: tandrii <tandrii@google.com>
Date: Wed Aug 17 15:47:47 2016

The following revision refers to this bug:
  https://chrome-internal.googlesource.com/infra/infra_internal.git/+/9296412d28cfc812a8ab8664d1d47a0545603a11

commit 9296412d28cfc812a8ab8664d1d47a0545603a11
Author: tandrii <tandrii@google.com>
Date: Wed Aug 17 17:12:00 2016

The following revision refers to this bug:
  https://chrome-internal.googlesource.com/infra/infra_internal.git/+/42524e8ed8797f8105ee45bf1dda1bda5d86f779

commit 42524e8ed8797f8105ee45bf1dda1bda5d86f779
Author: tandrii <tandrii@google.com>
Date: Wed Aug 17 18:25:25 2016

 Issue 640252  has been merged into this issue.

See  issue 640252 : this isn't so simple. If there are dependent CLs, all but the first one would be marked by Gerrit as "unmergable", even though from CQ/tryjob PoV they are totally fine.

So, I'm reverting my change here. A smarter way is needed and unless someone finds an easy way, IMO not worth the cost savings.

The following revision refers to this bug:
  https://chrome-internal.googlesource.com/infra/infra_internal.git/+/31ecb15f27bace2ee99e7bad1a55387f4d2c9d0f

commit 31ecb15f27bace2ee99e7bad1a55387f4d2c9d0f
Author: tandrii <tandrii@google.com>
Date: Tue Aug 23 19:32:48 2016

I've looked at REST api of Gerrit: CQ can't rely on merge itself. There is https://review.openstack.org/Documentation/rest-api-changes.html#get-related-changes  and it might be possible to figure it out, but nowhere trivial.

We might eventually end up with some kind of auto-bot that would verify this, but for now, it's Pri3 or wontfix.