WebContentsObserver::RenderFrameCreated should be called for every RenderFrameHost.

WebContentsObserver::RenderFrameCreated should be called for every RenderFrameHost.

BUG=364974

[ncarter (slow), 2014-04-22 21:44:38]

Nasko or John -- please have a look. I'm not at all certain of this approach,
but I think this is the smallest change we could make that ensures that
RenderFrameCreated is always called.

We could call it deeper in the stack (say, just after
RenderFrameHostFactory::Create) but my thinking is that observer calls should
happen after we've finished making mutations to the frame tree and the RFHM.

[jam, 2014-04-22 23:07:31]

lgtm

[Ken Russell (switch to Gerrit), 2014-05-12 22:45:01]

The linux_gpu_triggered_tests failures look real, and caused by this patch.
Please investigate them before committing.

[ncarter (slow), 2014-05-12 23:09:48]

On 2014/05/12 22:45:01, Ken Russell wrote:
> The linux_gpu_triggered_tests failures look real, and caused by this patch.
> Please investigate them before committing.

Thanks for the drive-by comment Ken. I was assuming as such, but I've had so
many dirty try runs it was tough to tell.

[ncarter (slow), 2014-05-12 23:11:01]

On 2014/05/12 23:09:48, ncarter wrote:
> On 2014/05/12 22:45:01, Ken Russell wrote:
> > The linux_gpu_triggered_tests failures look real, and caused by this patch.
> > Please investigate them before committing.
> 
> Thanks for the drive-by comment Ken. I was assuming as such, but I've had so
> many dirty try runs it was tough to tell.

(btw this patch is not currently lgtmed; it's evolved a great deal since patch
set 2 I'll be getting a de novo review before it lands)

[Ken Russell (switch to Gerrit), 2014-05-12 23:53:00]

On 2014/05/12 23:11:01, ncarter wrote:
> On 2014/05/12 23:09:48, ncarter wrote:
> > On 2014/05/12 22:45:01, Ken Russell wrote:
> > > The linux_gpu_triggered_tests failures look real, and caused by this
patch.
> > > Please investigate them before committing.
> > 
> > Thanks for the drive-by comment Ken. I was assuming as such, but I've had so
> > many dirty try runs it was tough to tell.
> 
> (btw this patch is not currently lgtmed; it's evolved a great deal since patch
> set 2 I'll be getting a de novo review before it lands)

Sounds good, and thanks for your understanding. :) As you may have seen, the GPU
bots are on the verge of becoming required steps on the CQ, so I'm particularly
concerned about breakage on them in the next few days.

[ncarter (slow), 2014-05-22 21:05:17]

On 2014/05/12 23:53:00, Ken Russell wrote:
> On 2014/05/12 23:11:01, ncarter wrote:
> > On 2014/05/12 23:09:48, ncarter wrote:
> > > On 2014/05/12 22:45:01, Ken Russell wrote:
> > > > The linux_gpu_triggered_tests failures look real, and caused by this
> patch.
> > > > Please investigate them before committing.
> > > 
> > > Thanks for the drive-by comment Ken. I was assuming as such, but I've had
so
> > > many dirty try runs it was tough to tell.
> > 
> > (btw this patch is not currently lgtmed; it's evolved a great deal since
patch
> > set 2 I'll be getting a de novo review before it lands)
> 
> Sounds good, and thanks for your understanding. :) As you may have seen, the
GPU
> bots are on the verge of becoming required steps on the CQ, so I'm
particularly
> concerned about breakage on them in the next few days.

I'm moving this review to https://codereview.chromium.org/292363003