Reimplement vsync latency UMA to be based on BeginImplFrame rather than BeginMainFrame

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 ad8770af97ccd0cf399c05f16a8f41797a8dd511..70a69cafb727acdec3bfdfdfedf02beb066c2f4a 100644
--- a/tools/metrics/histograms/histograms.xml
+++ b/tools/metrics/histograms/histograms.xml
@@ -60657,11 +60657,11 @@ http://cs/file:chrome/histograms.xml - but prefer this file for new entries.
   <summary>How long it takes the swap ack to return after a swap.</summary>
 </histogram>
 
-<histogram name="Scheduling.VsyncToBeginMainFrameLatency2" units="microseconds">
+<histogram name="Scheduling.VsyncToBeginImplFrameLatency2" units="microseconds">
   <owner>stanisc@chromium.org</owner>
   <summary>
     The time from v-sync to when the main side actually starts the
-    BeginMainFrame.
+    BeginImplFrame.
   </summary>
 </histogram>
 
@@ -115253,7 +115253,7 @@ from previous Chrome versions.
   <affected-histogram name="Scheduling.PrepareTilesDuration2"/>
   <affected-histogram name="Scheduling.SwapAckWasFast"/>
   <affected-histogram name="Scheduling.SwapToAckLatency2"/>
-  <affected-histogram name="Scheduling.VsyncToBeginMainFrameLatency2"/>
+  <affected-histogram name="Scheduling.VsyncToBeginImplFrameLatency2"/>
 </histogram_suffixes>
 
 <histogram_suffixes name="CompositorTimingHistoryProcess2" separator="."