Enable SiteEngagementEvictionPolicy behind --enable-site-eviction-policy flag.

chrome/browser/engagement/site_engagement_service.cc

Index: chrome/browser/engagement/site_engagement_service.cc
diff --git a/chrome/browser/engagement/site_engagement_service.cc b/chrome/browser/engagement/site_engagement_service.cc
index b4e29444b6ea627e489b907bd8ed6e6d735afd37..b03d4d7519fd25a1c44ea836360db08155aca0e5 100644
--- a/chrome/browser/engagement/site_engagement_service.cc
+++ b/chrome/browser/engagement/site_engagement_service.cc
@@ -11,6 +11,7 @@
 #include "base/time/clock.h"
 #include "base/values.h"
 #include "chrome/browser/content_settings/host_content_settings_map_factory.h"
+#include "chrome/browser/engagement/site_engagement_eviction_policy.h"
 #include "chrome/browser/engagement/site_engagement_helper.h"
 #include "chrome/browser/engagement/site_engagement_service_factory.h"
 #include "chrome/common/chrome_switches.h"
@@ -157,7 +158,8 @@ SiteEngagementService* SiteEngagementService::Get(Profile* profile) {
 // static
 bool SiteEngagementService::IsEnabled() {
   return base::CommandLine::ForCurrentProcess()->HasSwitch(
-      switches::kEnableSiteEngagementService);
+             switches::kEnableSiteEngagementService) ||

[raymes, 2015/09/24 01:55:30]

Hmm - should we just get rid of this switch? Do you see a good use for it? Is it
just for the chrome:// page?

[calamity, 2015/10/26 03:01:57]

We definitely want this switch around for experiments. I was thinking running
with control, engagement service only, and engagement service with eviction
groups. That would give us a better idea of the base cost of the engagement
service.

This is already done in the rebase actually.

+         SiteEngagementEvictionPolicy::IsEnabled();
 }
 
 SiteEngagementService::SiteEngagementService(Profile* profile)