Add three new histogram to tracking QUIC packet reordering:

tools/metrics/histograms/histograms.xml

Index: tools/metrics/histograms/histograms.xml
diff --git a/tools/metrics/histograms/histograms.xml b/tools/metrics/histograms/histograms.xml
index 357e28c2ca987420d18ed4cb081f574bcdd0bb0c..34f90aac71c47ae804f4df0ffcfd4e7ee2ebc476 100644
--- a/tools/metrics/histograms/histograms.xml
+++ b/tools/metrics/histograms/histograms.xml
@@ -13546,6 +13546,28 @@ Therefore, the affected-histogram name has to have at least one dot in it.
   </summary>
 </histogram>
 
+<histogram name="Net.QuicSession.MaxReordering">
+  <owner>rch@chromium.org</owner>
+  <summary>
+    The maximum packet sequence number reordering observed by a QUIC connection.
+  </summary>
+</histogram>
+
+<histogram name="Net.QuicSession.MaxReorderingTime" units="percent">
+  <owner>rch@chromium.org</owner>
+  <summary>
+    The ratio of the maximum reordering time of a QUIC packet to the min rtt.
+  </summary>
+</histogram>
+
+<histogram name="Net.QuicSession.MaxReorderingTimeLongRtt" units="percent">
+  <owner>rch@chromium.org</owner>
+  <summary>
+    The ratio of the maximum reordering time of a QUIC packet to the min rtt,

[jar (doing other things), 2014/05/23 16:24:49]

nit: to reduce replication of prose, we often use the "field_trial" construct in
this file to suggest "alternate continuations of the histogram name" as well as
the associated "additional prose."  You named the histograms perfectly for using
this pattern.

In this case, with such a small amount of prose... and only one replication, it
is borderline... but it is often the case that the list grows, and having the
field trial just simplifies things.

In this case, it is fifty-fifty about which is better... and this is done... so
I'll LGTM.

+    only for those sessions with a min rtt larger than 100 ms.
+  </summary>
+</histogram>
+
 <histogram name="Net.QuicSession.NumOpenStreams">
   <owner>rch@chromium.org</owner>
   <summary>