Enable QUIC for proxies based on Finch config and command line switch.

chrome/browser/io_thread.cc

Index: chrome/browser/io_thread.cc
diff --git a/chrome/browser/io_thread.cc b/chrome/browser/io_thread.cc
index 9ef53eeac4556f6865d2b244199b8d94a453161e..7e649cb7b8d09e5b1ff5c3d3a9d5718e78dc492d 100644
--- a/chrome/browser/io_thread.cc
+++ b/chrome/browser/io_thread.cc
@@ -1226,7 +1226,9 @@ bool IOThread::ShouldEnableQuic(const base::CommandLine& command_line,
     return true;
 
   return quic_trial_group.starts_with(kQuicFieldTrialEnabledGroupName) ||
-      quic_trial_group.starts_with(kQuicFieldTrialHttpsEnabledGroupName);
+      quic_trial_group.starts_with(kQuicFieldTrialHttpsEnabledGroupName) ||
+      base::FieldTrialList::FindFullName(
+          "DataReductionProxyUseQUICAsPrimaryOrigin") == "Enabled";

[Ryan Hamilton, 2015/02/06 22:23:52]

Ah, I see. Hm, I'm not super in love this this. This allows the volume of QUIC
traffic to be turned up it two different studies; only one of which is
controlled by quic-dev (and hence is hooked into traffic-teams plans).

You probably proposed this in your initial email I didn't grok it, but...
Perhaps we should split the existing enable_quic support into
enable_quic_to_origin and enable_quic_to_proxies. Your experiment could turn on
the latter but not the former. Would something like that seem reasonable? 

 }
 
 bool IOThread::ShouldEnableQuicPortSelection(