Add histograms for data usage

chrome/browser/android/data_usage/external_data_use_observer_bridge.cc

Index: chrome/browser/android/data_usage/external_data_use_observer_bridge.cc
diff --git a/chrome/browser/android/data_usage/external_data_use_observer_bridge.cc b/chrome/browser/android/data_usage/external_data_use_observer_bridge.cc
index 71ad96c18be6d3cd2446dc2be470b42e046d6195..837eb10f4be92a7c1b817d2da3ad57df78cbc769 100644
--- a/chrome/browser/android/data_usage/external_data_use_observer_bridge.cc
+++ b/chrome/browser/android/data_usage/external_data_use_observer_bridge.cc
@@ -119,6 +119,10 @@ void ExternalDataUseObserverBridge::FetchMatchingRulesDone(
 
   data_use_tab_model_->RegisterURLRegexes(
       app_package_name_native, domain_path_regex_native, label_native);
+
+  io_task_runner_->PostTask(
+      FROM_HERE, base::Bind(&ExternalDataUseObserver::OnFetchMatchingRulesDone,

[tbansal1, 2016/01/13 18:06:52]

What are we trying to measure here? The way it is instrumented now, it will also
take into account time taken for 2 thread hops, which is not useful.

May be the purpose is to measure the time it takes since the start of Chrome to
until the matching rules are first fetched (because all bytes during that time
would be lost -- and we want to know that)?

I guess after the first fetch, it is not important how much time the subsequent
fetches take (e.g., even if they take 1 minute, we are still okay because rules
are valid for much longer duration).

[Raj, 2016/01/13 23:33:51]

Agreed. I guess duration for first fetch is more valuable, than for subsequent
fetches.
But I thought the thread hops are negligible, and adding up those times don't
matter? 

+                            external_data_use_observer_));
 }
 
 void ExternalDataUseObserverBridge::ReportDataUse(