Implement NavigationThrottle::BLOCK_REQUEST_AND_COLLAPSE.

content/public/browser/navigation_throttle.h

Index: content/public/browser/navigation_throttle.h
diff --git a/content/public/browser/navigation_throttle.h b/content/public/browser/navigation_throttle.h
index ba49aa73a716aa81af9106d7b875df3f18fa26ac..76391291df806961d9e4ad56641014ad15c61256 100644
--- a/content/public/browser/navigation_throttle.h
+++ b/content/public/browser/navigation_throttle.h
@@ -39,6 +39,11 @@ class CONTENT_EXPORT NavigationThrottle {
     // is navigated.
     BLOCK_REQUEST,
 
+    // Blocks a navigation taking place in a subframe, and collapses the frame
+    // owner element in the parent frame (i.e. removes it from the layout). Can
+    // be be returned from WillStartRequest or WillRedirectRequest.
+    BLOCK_REQUEST_AND_COLLAPSE,

[clamy, 2017/01/18 15:10:55]

I'm wondering whether we should add a new ThrottleCheckResult vs providing a
method on the NavigationHandle or the NavigationThrottle to collapse the frame.
The NavigationThrottle would return an existing result in that case.

[engedy, 2017/01/18 23:37:14]

I am fine with either :-). @Nasko, WDYT?

[engedy, 2017/02/17 18:09:25]

On second thought, I think it makes for a better and harder-to-misuse API to
bundle collapsing with blocking the navigation. Collapsing without blocking
makes little sense to me, and would just waste of resources.

+
     // Blocks a navigation due to rules asserted by a response (for instance,
     // embedding restrictions like 'X-Frame-Options'). This result will only
     // be returned from WillProcessResponse.