Add startup metrics that measure the performance of the first web contents.

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 e22ab98ff6c812e948a41928b1116d88fe0b7577..7a68d980730e4e05b42f67319bf0906ea2e796d8 100644
--- a/tools/metrics/histograms/histograms.xml
+++ b/tools/metrics/histograms/histograms.xml
@@ -35391,6 +35391,52 @@ Therefore, the affected-histogram name has to have at least one dot in it.
   </summary>
 </histogram>
 
+<histogram
+    name="Startup.Experimental.FirstWebContents.MainFrameLoad.ManyBuckets"

[Alexei Svitkine (slow), 2014/12/03 20:05:57]

Honestly, I'd just remove the Experimental substring here - it just makes it
longer without much benefit.

Also, if we decide to keep one of these metrics, we won't need to rename them,
then.

[erikchen, 2014/12/03 21:38:41]

I would prefer to keep the names as is, since I intend to rename the metric
regardless, and this makes it more clear that these are temporary metrics.

Regardless of the final metric, I intend to remove the final affix from the
name, since "ManyBuckets" and "StandardBuckets" add no meaning to the metric.

If there's some hidden cost that I'm no aware of to orphaning histograms, let me
know and I'll happily make the change.

[Alexei Svitkine (slow), 2014/12/03 22:09:33]

The cost is keeping the old entries in the XML file and marking them as
obsolete. (Since we want to be able to decode even old data.)

[Ilya Sherman, 2014/12/03 22:16:11]

Alternately, the cost is losing continuity for your data -- though if you know
where to look, it's easy enough to just view both histograms side-by-side on the
dashboard.

[erikchen, 2014/12/03 22:20:11]

Understood.

Since I plan on renaming the metric anyways to remove the final affix, I will
continue with my plan to not rename this temporary histogram.

+    units="milliseconds">
+  <owner>erikchen@chromium.org</owner>
+  <summary>
+    An experimental metric only collected on the dev and canary channels.
+    Measure the elapsed time from process launch to the first main frame load of
+    the first web contents. Uses significantly more buckets, with reduced
+    ranges.
+  </summary>
+</histogram>
+
+<histogram
+    name="Startup.Experimental.FirstWebContents.MainFrameLoad.StandardBuckets"
+    units="milliseconds">
+  <owner>erikchen@chromium.org</owner>
+  <summary>
+    An experimental metric only collected on the dev and canary channels.
+    Measure the elapsed time from process launch to the first main frame load of
+    the first web contents. Uses standard bucket ranges.
+  </summary>
+</histogram>
+
+<histogram
+    name="Startup.Experimental.FirstWebContents.NonEmptyPaint.ManyBuckets"
+    units="milliseconds">
+  <owner>erikchen@chromium.org</owner>
+  <summary>
+    An experimental metric only collected on the dev and canary channels.
+    Measure the elapsed time from process launch to the first non-empty paint of
+    the first web contents. Uses significantly more buckets, with reduced
+    ranges.
+  </summary>
+</histogram>
+
+<histogram
+    name="Startup.Experimental.FirstWebContents.NonEmptyPaint.StandardBuckets"
+    units="milliseconds">
+  <owner>erikchen@chromium.org</owner>
+  <summary>
+    An experimental metric only collected on the dev and canary channels.
+    Measure the elapsed time from process launch to the first non-empty paint of
+    the first web contents. Uses standard bucket ranges.
+  </summary>
+</histogram>
+
 <histogram name="Startup.Fling.TimeToDisplayVideo" units="milliseconds">
   <owner>jeremy@chromium.org</owner>
   <summary>