[ProtocolHandlers] Add a permission bubble client for RPH requests.

chrome/browser/ui/browser.cc

Index: chrome/browser/ui/browser.cc
diff --git a/chrome/browser/ui/browser.cc b/chrome/browser/ui/browser.cc
index 46a8e6495b000cf5aa89635ad2310fac9c224019..858522500b4bb47f0c1ee4a861ed8fb176a51062 100644
--- a/chrome/browser/ui/browser.cc
+++ b/chrome/browser/ui/browser.cc
@@ -1622,8 +1622,7 @@ void Browser::RegisterProtocolHandler(WebContents* web_contents,
     window_->GetLocationBar()->UpdateContentSettingsIcons();
   }
 
-  RegisterProtocolHandlerInfoBarDelegate::Create(
-      InfoBarService::FromWebContents(web_contents), registry, handler);
+  registry->RequestPermission(web_contents, handler);

[koz (OOO until 15th September), 2014/02/06 03:03:26]

Better to not add RequestPermission() to ProtocolHandlerRegistry, because it
doesn't handle view level concerns like that. I think you should inline that
function here.

[Greg Billock, 2014/02/06 22:23:49]

This is intended to contain the bubble/infobar split within the protocol handler
registry rather than making it in browser.cc. Have a look: if you want I can
move the whole thing here.

[koz (OOO until 15th September), 2014/02/06 23:34:10]

Yes please. I think it would make more sense to have this logic in browser.cc.
ProtocolHandlerRegistry is currently completely ignorant of UI concepts like
WebContents and the difference between when it's appropriate to show an infobar
or a bubble, whereas browser.cc is a view class and deals with this kind of
thing already - the class creates infobars in many other situations.

[Greg Billock, 2014/02/10 23:12:51]

Sounds good.

The context is that I've been doing several other conversions, and by and large
the feature-specific coordinator class is what triggers the infobar in these
cases (webrtc, midi, quota, desktop notifications, multiple downloads,
geolocation all have this setup), so that's the expectation I'm seeing. That
said, I don't think it's that big a thing.

[koz (OOO until 15th September), 2014/02/11 02:17:07]

Yeah, it's probably not, but I'd be afraid of more UI things slipping into
ProtocolHandlerRegistry. Thanks!

 }
 
 void Browser::UpdatePreferredSize(WebContents* source,