Add Precache.Fetch.MinWeight UMA.

components/precache/content/precache_manager.cc

Index: components/precache/content/precache_manager.cc
diff --git a/components/precache/content/precache_manager.cc b/components/precache/content/precache_manager.cc
index 7732807e8f8e6af204265c69fd075a3f9a0c9bc7..6f64cb9ba9497d13682aba71ed0292f5b66dc6c6 100644
--- a/components/precache/content/precache_manager.cc
+++ b/components/precache/content/precache_manager.cc
@@ -14,6 +14,7 @@
 #include "base/memory/ref_counted.h"
 #include "base/metrics/field_trial.h"
 #include "base/metrics/histogram_macros.h"
+#include "base/optional.h"
 #include "base/strings/string_number_conversions.h"
 #include "base/strings/string_util.h"
 #include "base/time/time.h"
@@ -272,7 +273,7 @@ void PrecacheManager::OnGetUnfinishedWorkDone(
           base::TimeDelta::FromHours(6)) {
     PrecacheFetcher::RecordCompletionStatistics(
         *unfinished_work, unfinished_work->top_host_size(),
-        unfinished_work->resource_size());
+        unfinished_work->resource_size(), base::nullopt);

[jamartin, 2017/02/22 22:54:07]

When does this case happen? When we haven't precached everything that we should
nor reached any of the caps yet the precaching session has lasted at least 6
hours? If yes, I wonder if we should still report the min_weight, since it is
still representative of the last resource precached by this client.

[twifkak, 2017/02/23 23:28:41]

Yes. The problem is that the weights aren't in the unfinished_work (just the
weight_ratios), so I'd need to add an API hook and some extra work to compute
that. I presume that the fraction of precaches that hit this timeout is tiny
enough that this isn't worth the bother.

     unfinished_work.reset(new PrecacheUnfinishedWork);
   }
   // If this prefetch is new, set the start time.