Don't force full invalidation on box-sizing change

LayoutTests/fast/repaint/box-sizing-expected.txt

Index: LayoutTests/fast/repaint/box-sizing-expected.txt
diff --git a/LayoutTests/compositing/layer-creation/compositing-reason-removed-expected.txt b/LayoutTests/fast/repaint/box-sizing-expected.txt
similarity index 54%
copy from LayoutTests/compositing/layer-creation/compositing-reason-removed-expected.txt
copy to LayoutTests/fast/repaint/box-sizing-expected.txt
index f7b730dca491caae37c1779ca75afc492cc4718a..f091706c1bb9305757d81a74f712c5f7c132850e 100644
--- a/LayoutTests/compositing/layer-creation/compositing-reason-removed-expected.txt
+++ b/LayoutTests/fast/repaint/box-sizing-expected.txt
@@ -6,7 +6,10 @@
       (contentsOpaque 1)
       (drawsContent 1)
       (repaint rects
-        (rect 8.00 8.00 30.00 30.00)
+        (rect 400.00 0.00 40.00 140.00)
+        (rect 380.00 0.00 20.00 140.00)
+        (rect 0.00 100.00 440.00 40.00)
+        (rect 0.00 80.00 440.00 20.00)

[esprehn, 2014/06/20 08:15:15]

Why do we issue 4 invalidations instead of just 1 now?

[Xianzhu, 2014/06/20 16:06:17]

This is a new test. There is no original file actually, but codereview/git
selected some similar file as the original file :)

       )
     )
   )