Refactor root element background painting

LayoutTests/platform/android/tables/mozilla/bugs/bug2479-2-expected.txt

Index: LayoutTests/platform/android/tables/mozilla/bugs/bug2479-2-expected.txt
diff --git a/LayoutTests/platform/android/tables/mozilla/bugs/bug2479-2-expected.txt b/LayoutTests/platform/android/tables/mozilla/bugs/bug2479-2-expected.txt
index 3ee7c3e13a309aecaaae27d35a3e07da39fa774f..71cf5f48b02b7a5d09ebe81a6c1e0aa1a79cb104 100644
--- a/LayoutTests/platform/android/tables/mozilla/bugs/bug2479-2-expected.txt
+++ b/LayoutTests/platform/android/tables/mozilla/bugs/bug2479-2-expected.txt
@@ -1,7 +1,7 @@
 layer at (0,0) size 800x600
   LayoutView at (0,0) size 800x600
 layer at (0,0) size 800x551
-  LayoutBlockFlow {HTML} at (0,0) size 800x551 [bgcolor=#00FF99]
+  LayoutBlockFlow {HTML} at (0,0) size 800x551

[chrishtr, 2015/06/09 21:07:52]

ditto

[trchen, 2015/06/09 23:36:07]

This is due to the way we dump renderer tree.

The LayoutView is supposed to report background color, but somehow we don't
report background color if the renderer has no parent. (bug or feature?)

Whether the root element renderer reports background color isn't all that
important because it no longer paints background. It is not reported because it
has the same background color as its parent (the LayoutView), which doesn't make
sense at all (background color is not inherited, and having the same background
color as the parent doesn't imply painting can be skipped).

This type counts maybe 40% of the rebaseline.

I think the behavior is very buggy and we should fix LayoutTreeAsText, but it is
going to need hundreds or rebaseline, so definitely not in this CL.

[chrishtr, 2015/06/10 01:03:39]

Good analysis, cool. Please file a bug to track this also.

[trchen, 2015/06/10 01:41:05]

crbug.com/498592

     LayoutBlockFlow {BODY} at (57,64) size 686x423 [bgcolor=#99FFCC] [border: (3px solid #000000)]
       LayoutBlockFlow {P} at (3,3) size 680x64
         LayoutText {#text} at (0,7) size 652x50