Proper unicode-range font loading behavior

Proper unicode-range font loading behavior

Remote fonts defined by @font-face rules should be loaded only if any
characters in its unicode-range descriptor are used [1].

Before this patch, remote fonts are loaded when the first glyph page for
the font is created. It did not 100% prevent unnecessary downloads, as
(1) All remote fonts in the fallback list were loaded, and (2) Use of
codepoint near the unicode-range could trigger load.

This patch adds m_customFontToLoad field to GlyphPage. It contains
CustomFontData for the first unloaded remote font in the fallback list.
GlyphPage::glyphDataForCharacter() triggers the font load.

text-intro-09-b.svg had to be changed to wait loads because SVGHebrew
and 'Ezra SIL SR' are now loaded sequentially, not at once.

[1] http://www.w3.org/TR/css3-fonts/#composite-fonts

BUG=247920
TEST=blink_platform_unittests,fast/css/font-face-unicode-range-overlap-load.html

Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=172443

[eae, 2014-04-23 18:38:38]

LGTM

https://codereview.chromium.org/243453003/diff/110001/Source/platform/fonts/G...
File Source/platform/fonts/GlyphPageTreeNode.cpp (right):

https://codereview.chromium.org/243453003/diff/110001/Source/platform/fonts/G...
Source/platform/fonts/GlyphPageTreeNode.cpp:230: for (int j = from; j < to; j++)
{
Here and elsewhere shouldn't j (and from) be unsigned?

[Kunihiko Sakamoto, 2014-04-24 01:49:24]

Thank you for review.

https://codereview.chromium.org/243453003/diff/110001/Source/platform/fonts/G...
File Source/platform/fonts/GlyphPageTreeNode.cpp (right):

https://codereview.chromium.org/243453003/diff/110001/Source/platform/fonts/G...
Source/platform/fonts/GlyphPageTreeNode.cpp:230: for (int j = from; j < to; j++)
{
On 2014/04/23 18:38:39, eae wrote:
> Here and elsewhere shouldn't j (and from) be unsigned?

I'd prefer to keep them int, as calculation of |from| and |to| may involve
negative numbers. Also, range.from() and range.to() are UChar32, which is
int32_t, so static_cast somewhere is inevitable.

[Kunihiko Sakamoto, 2014-04-24 01:50:20]

The CQ bit was checked by ksakamoto@chromium.org

[commit-bot: I haz the power, 2014-04-24 01:50:38]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/ksakamoto@chromium.org/243453003/110001

[commit-bot: I haz the power, 2014-04-24 03:05:46]

Change committed as 172443