Gesture event hit test refactoring and reduction

LayoutTests/fast/events/hit-test-counts-expected.txt

Index: LayoutTests/fast/events/hit-test-counts-expected.txt
diff --git a/LayoutTests/fast/events/hit-test-counts-expected.txt b/LayoutTests/fast/events/hit-test-counts-expected.txt
index f7d0d73676701f5b2f1a3215fbb667c527f320fb..9b81dd72fe67151571a344e4ae7875d9417bad6d 100644
--- a/LayoutTests/fast/events/hit-test-counts-expected.txt
+++ b/LayoutTests/fast/events/hit-test-counts-expected.txt
@@ -16,11 +16,11 @@ Wheel: 1
 TouchStart: 1
 TouchMove: 0
 TouchEnd: 0
-GestureTapDown: 3
-GestureShowPress: 5
-GestureTap: 7
-GestureScrollBegin: 2
-GestureTapCancel: 2
+GestureTapDown: 1
+GestureShowPress: 2
+GestureTap: 6
+GestureScrollBegin: 1
+GestureTapCancel: 1
 GestureScrollUpdate: 0
 GestureScrollEnd: 0
 
@@ -34,11 +34,11 @@ Wheel: 1 1 1
 TouchStart: 1 1 1
 TouchMove: 0 0 0
 TouchEnd: 0 0 0
-GestureTapDown: 5 5 5
-GestureShowPress: 7 7 7
-GestureTap: 3 3 7
-GestureScrollBegin: 4 4 4
-GestureTapCancel: 2 2 2
+GestureTapDown: 1 1 1
+GestureShowPress: 2 2 2
+GestureTap: 2 2 6
+GestureScrollBegin: 1 1 1
+GestureTapCancel: 1 1 1
 GestureScrollUpdate: 0 0 0
 GestureScrollEnd: 0 0 0
 
@@ -52,11 +52,11 @@ Wheel: 1 1 0
 TouchStart: 1 1 0
 TouchMove: 0 0 0
 TouchEnd: 0 0 0
-GestureTapDown: 5 14 50
-GestureShowPress: 7 19 67
-GestureTap: 3 7 0
-GestureScrollBegin: 3 3 0
-GestureTapCancel: 2 2 0
+GestureTapDown: 1 4 16
+GestureShowPress: 2 8 32

[esprehn, 2014/06/27 08:33:58]

woah, we did 67 hit tests before?

[Rick Byers, 2014/06/27 15:38:23]

Worse! Total of 162 for the TapDown and ShowPress and probably another big chunk
for the Tap but it's not showing here because EventSender currently always sends
a 0 radius for tap.  I'm fixing that separately.

This is the issue I found yesterday in the previous CL:
https://codereview.chromium.org/354613002/#msg3.  It's an edge case (a tap which
covers multiple nested iframes) that triggers a pathological hit-test explosion
in touch adjustment (which searches for the best node within the tap bounding
box).  It's why I added this third section to the test.  I could probably come
up with something with huge multi-second delays if I added a few more tiny
nested iframes :-).  Anyway this should be rare in practice, but I'll get it
down to something reasonable.

+GestureTap: 2 6 0
+GestureScrollBegin: 1 1 0
+GestureTapCancel: 1 1 0
 GestureScrollUpdate: 0 0 0
 GestureScrollEnd: 0 0 0