Move MimeTypeResourceHandler before ThrottlingResourceHandler

chrome/browser/renderer_host/safe_browsing_resource_throttle.cc

Index: chrome/browser/renderer_host/safe_browsing_resource_throttle.cc
diff --git a/chrome/browser/renderer_host/safe_browsing_resource_throttle.cc b/chrome/browser/renderer_host/safe_browsing_resource_throttle.cc
index d61f3cb3bc4e1d5afb37f65e749ac96b84b78e1c..8210b0b618fcc91d495c6c06698762f8dd758627 100644
--- a/chrome/browser/renderer_host/safe_browsing_resource_throttle.cc
+++ b/chrome/browser/renderer_host/safe_browsing_resource_throttle.cc
@@ -169,6 +169,10 @@ void SafeBrowsingResourceThrottle::WillProcessResponse(bool* defer) {
   }
 }
 
+bool SafeBrowsingResourceThrottle::MustProcessResponseBeforeReadingBody() {
+  return true;

[mmenke, 2016/08/19 16:33:02]

Also, if we have any SafeBrowsing integration tests for that pause requests in
WillProcessResponse, and then cancel them, we should have an integration test
where we go through that path for a cacheable resource, then change the
safebrowsing result to not block the resource, turn off the test server, and
re-send the request, making sure we don't get the cached result.

If we *don't* have any integration tests for that SafeBrowsing path, then we
really should, but adding them is well beyond the scope of this CL, and this
doesn't make test coverage any worse.

[mmenke, 2016/08/19 16:33:02]

Think this needs an explanatory comment.

[clamy, 2016/09/01 23:19:40]

Added a comment.

[clamy, 2016/09/01 23:19:40]

Considering that we now have a unit test to check for
MustProcessResponseBeforeReadingBody and that this CL is big enough, I have
marked adding a browser test as a TODO to do in a followup patch.

+}
+
 void SafeBrowsingResourceThrottle::WillRedirectRequest(
     const net::RedirectInfo& redirect_info,
     bool* defer) {