Use a composed tree based traversal in computing quads in LinkHighlight::computeQuads().

Use a composed tree based traversal in computing quads in LinkHighlight::computeQuads().

This is a follow-up patch for https://codereview.chromium.org/448593002 to fix a crash.

The previous patch uses renderer.slowFirstChild() to get an child renderer in computeQuads(). 
However, this is not always correct. For example, suppose that <a><div></div></a> is given, the render tree would be:

  RenderBlock (anonymous) 0x6110004a6400
    RenderInline 0x60c00014e6c0 A       0x60d00006cf10
  RenderBlock (anonymous) 0x6110004a6180
    RenderBlock 0x6110004a62c0  DIV     0x60b00004d930 CLASS="_muf"

In this case, there is no child renderer for RenderInline object(0x60c00014e6c0).

Instead, LinkHighlight::computeQuads() should use a composed tree based traversal to compute quads correctly because we've already used a composed tree based traversal to get an parent object in the previous patch.

BUG=417191
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=183756

[hayato, 2014-10-06 07:45:17]

WIP.
LinkHighlightTest.verifyWebViewImplIntegration seemed to fail. Let me take a
loot.

[hayato, 2014-10-06 08:28:01]

hayato@chromium.org changed reviewers:
+ dglazkov@chromium.org, esprehn@chromium.org

[hayato, 2014-10-06 08:28:27]

PTAL. The patch is ready for reviews.

[hayato, 2014-10-14 05:27:37]

hayato@chromium.org changed reviewers:
+ kochi@chromium.org

[hayato, 2014-10-15 04:33:59]

Friendly ping. I've fixed the falling layout test.
Could someone review this?

[dglazkov, 2014-10-15 15:47:53]

The CQ bit was checked by dglazkov@chromium.org

[dglazkov, 2014-10-15 15:47:54]

lgtm

[commit-bot: I haz the power, 2014-10-15 15:48:33]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/622253004/80001

[commit-bot: I haz the power, 2014-10-15 15:54:21]

Committed patchset #5 (id:80001) as 183756