The cause is unclear, but anyway I'm disabling the test.

Test output:

[ RUN      ] NavigationHandleImplBrowserTest.VerifyRequestContextTypeForFrameTree
[3956:5276:0227/155854.201:6288509:ERROR:interface_registry.cc(210)] Failed to locate a binder for interface: memory_instrumentation::mojom::Coordinator requested by: content_gpu exposed by: content_browser via InterfaceProviderSpec "service_manager:connector".
[1892:3500:0227/155854.202:6288509:INFO:media_foundation_video_encode_accelerator_win.cc(329)] Windows versions earlier than 8 are not supported.
[1892:3500:0227/155854.203:6288509:INFO:media_foundation_video_encode_accelerator_win.cc(329)] Windows versions earlier than 8 are not supported.
[3956:5276:0227/155854.229:6288540:ERROR:interface_registry.cc(210)] Failed to locate a binder for interface: memory_instrumentation::mojom::Coordinator requested by: content_renderer exposed by: content_browser via InterfaceProviderSpec "service_manager:connector".
e:\c\win\src\contentrowserrame_host
avigation_handle_impl_browsertest.cc(774): error: Expected: (previous_throttle) != (installer.navigation_throttle()), actual: 0DB4EAA8 vs 0DB4EAA8
[  FAILED  ] NavigationHandleImplBrowserTest.VerifyRequestContextTypeForFrameTree, where TypeParam =  and GetParam() =  (327 ms)

Disabling: https://codereview.chromium.org/2726503002

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/b56108f35966114a1f0423c684c1e942d18f5511

commit b56108f35966114a1f0423c684c1e942d18f5511
Author: yutak <yutak@chromium.org>
Date: Tue Feb 28 09:24:53 2017

Disable VerifyRequestContextTypeForFrameTree on Windows.

BUG= 696915 
TBR=fsamuel@chromium.org,sergeyu@chromium.org,dvadym@chromium.org
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_site_isolation

Review-Url: https://codereview.chromium.org/2726503002
Cr-Commit-Position: refs/heads/master@{#453551}

[modify] https://crrev.com/b56108f35966114a1f0423c684c1e942d18f5511/content/browser/frame_host/navigation_handle_impl_browsertest.cc

I'm not quite sure who to assign this bug to...

carlosk (test author): Do you have any idea?

According to my local testing, this failure reproduces about 2% of times
on Debug bulid.

This test was already disabled. Assigning to carlosk for triage, and removing Sheriff label.

Starting to take a look. Lowering priority as the test is already disabled.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/87331d860f547f0a5b5df862b77e7a353ea2724e

commit 87331d860f547f0a5b5df862b77e7a353ea2724e
Author: carlosk <carlosk@chromium.org>
Date: Thu Mar 02 18:43:41 2017

Fix NavigationHandleImplBrowserTest flakiness.

NavigationHandleImplBrowserTest.VerifyRequestContextTypeForFrameTree has been
flaky because it relied on pointer comparisons for checking that new throttles
were being properly created for each new navigation. This is an unsafe solution
and has already caused problems in the past as it's not uncommon that new
instances of the same type of object are created at precisely the same memory
address. This is especially the case on Windows, where this test has been flaky.

This change uses a throttle-install-counter instead that works just as well for
(indirectly) checking that new throttles were installed without the pointer
comparison drawbacks. It also re-enables that test on Windows.

Moreover it moves class member default initializations to be inline with their
declarations to simplify the code and cut back on line count.

BUG= 696915 
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_site_isolation

Review-Url: https://codereview.chromium.org/2728783003
Cr-Commit-Position: refs/heads/master@{#454318}

[modify] https://crrev.com/87331d860f547f0a5b5df862b77e7a353ea2724e/content/browser/frame_host/navigation_handle_impl_browsertest.cc