Merges two WK fixes for RenderLayer::backgroundIsKnownToBeOpaque.

LayoutTests/compositing/contents-opaque/visibility-hidden-expected.txt

Index: LayoutTests/compositing/contents-opaque/visibility-hidden-expected.txt
diff --git a/LayoutTests/compositing/clip-child-by-non-stacking-ancestor-expected.txt b/LayoutTests/compositing/contents-opaque/visibility-hidden-expected.txt
similarity index 75%
copy from LayoutTests/compositing/clip-child-by-non-stacking-ancestor-expected.txt
copy to LayoutTests/compositing/contents-opaque/visibility-hidden-expected.txt
index caeddd2218da26ed3494e0b6bab420e2578e22d2..4f6bf93d4f25ffd658e5bde856d64be901f01edf 100644
--- a/LayoutTests/compositing/clip-child-by-non-stacking-ancestor-expected.txt
+++ b/LayoutTests/compositing/contents-opaque/visibility-hidden-expected.txt
@@ -7,9 +7,8 @@
       (drawsContent 1)
       (children 1
         (GraphicsLayer
-          (position 8.00 8.00)
-          (bounds 100.00 100.00)
-          (contentsOpaque 1)
+          (position 8.00 20.00)
+          (bounds 200.00 50.00)
           (drawsContent 1)
           (backgroundColor #C0C0C0)

[Xianzhu, 2015/05/07 15:51:05]

alokp@ this looks incorrect to me. The object has been set 'visibility: hidden',
why there is still a layer with background color?

[alokp, 2015/05/07 17:00:59]

May be assumptions have changed in 2 years since this patch landed.

What are you asking?
- Why is there even a layer for invisible element, or
- Why does this layer have a background color?

The important thing that this patch was trying to accomplish is to make sure we
do not mark layers as opaque when they are NOT. We were trying to be
conservative about it.

[Xianzhu, 2015/05/07 17:13:24]

I'm asking mainly the first question. Will contentsOpaque affect whether cc to
display the layer?

         )