fix the rendering error for drawPosText_asPaths

fix the rendering error for drawPosText_asPaths

A big thick-stroke text(>256*256, strokeWidth>0) is drawn on the basis of canonical text in drawPosText_asPaths. This 
behavior is the same with drawText_asPaths. So the textSize as well as strokeWidth in drawPosText_asPaths should be set 
as those in drawText_asPaths. After SkGlyphCache has been detached and constructed, the SkPaint should be set to 
kFill_Style to draw the correct path for the text, as that in drawText_asPaths.

In addtion, PathEffect installed in SkPaint should be considered, who affects the path generation for text. 
Current implementation of SkDraw::drawPosText_asPaths didn't consider this. As a result, big text and hairline stroked 
text with path effect is not correctly rendered by  SkCanvas::drawPosText. This patch also fixed this issue.

BUG=skia:2547, skia:2555
BUG=280221, 311731
@reed has landed a patch to fix the issues above: https://codereview.chromium.org/288983004/. So close this one.

[yunchao, 2014-05-14 15:48:50]

On 2014/05/14 08:31:09, Richard Ho wrote:

I will update expectations/gm/ignore-tests.txt if this CL is OK.

[zino, 2014-05-14 15:58:30]

@Richard,

I'm not reviewer but it seems that this is duplicated with my CL.
(https://codereview.chromium.org/273283003/)

[yunchao, 2014-05-14 16:23:57]

On 2014/05/14 15:58:30, zino wrote:

@zino,

I am aware of this bug from GM case stroketext several month ago, and fix it
recently standalone. When I search the potential crbug to link them into my CL,
I found somebody has indeed reported the same error in Cr issues, but I still
didn't notice your CL. 

Let reviewers to make a decision to land whose patch if it is OK. before that, I
want to say that: 
1) please don't setStrokeWidth as well as setStyle for SkPaint if it is
Haireline path for text drawing. Otherwise, it will lead to another bug.
2) update expectation/gm/ignore-tests.txt before you land a CL if your CL
changed gm case(s). Well, I suppose you have run unit tests and gm tests before
you submit a patch into upstream Skia.

[reed1, 2014-05-14 21:43:27]

... am reviewing

[yunchao, 2014-05-15 09:24:16]

On 2014/05/14 21:43:27, reed1 wrote:
> ... am reviewing

Hi, Mike and all, please take a look, I have updated this CL to take pathEffect
into consideration.

[yunchao, 2014-05-15 09:26:04]

On 2014/05/15 09:24:16, Richard Ho wrote:
> On 2014/05/14 21:43:27, reed1 wrote:
> > ... am reviewing
> 
> Hi, Mike and all, please take a look, I have updated this CL to take
pathEffect
> into consideration.

A GM case also added in another CL to ensure the rendering result of path effect
for big text and hairline stroked text:
https://codereview.chromium.org/290443004/

[yunchao, 2014-05-22 14:35:24]

On 2014/05/15 09:26:04, Richard Ho wrote:
> On 2014/05/15 09:24:16, Richard Ho wrote:
> > On 2014/05/14 21:43:27, reed1 wrote:
> > > ... am reviewing
> > 
> > Hi, Mike and all, please take a look, I have updated this CL to take
> pathEffect
> > into consideration.
> 
> A GM case also added in another CL to ensure the rendering result of path
effect
> for big text and hairline stroked text:
> https://codereview.chromium.org/290443004/

ping... PTAL, thanks!

[bungeman-skia, 2014-05-22 15:03:52]

See also https://codereview.chromium.org/20693003/ .

[reed1, 2014-05-22 15:28:55]

Hi.

Very sorry for the long delay in reviewing. We have had many fires to fight with
the recent Chrome milestone. I am now actively re-leaning some of the bugs, and
experimenting with your CL and others.

[reed1, 2014-05-22 15:29:30]

On 2014/05/22 15:28:55, reed1 wrote:
> Hi.
> 
> Very sorry for the long delay in reviewing. We have had many fires to fight
with
> the recent Chrome milestone. I am now actively re-leaning some of the bugs,
and
> experimenting with your CL and others.

https://codereview.chromium.org/20693003/

This is a very old CL that also tries to address some of the bugs. I am looking
at it as well...

[yunchao, 2014-05-22 16:00:21]

On 2014/05/22 15:28:55, reed1 wrote:
> Hi.
> 
> Very sorry for the long delay in reviewing. We have had many fires to fight
with
> the recent Chrome milestone. I am now actively re-leaning some of the bugs,
and
> experimenting with your CL and others.

Yeah, I know that. Heather told me that you are overloaded because of a very
important Chrome milestone(but I don't know the exact day, :) ). I hope you can
take a look when you have time.

[yunchao, 2014-05-22 16:04:03]

On 2014/05/22 16:00:21, Richard Ho wrote:
> On 2014/05/22 15:28:55, reed1 wrote:
> > Hi.
> > 
> > Very sorry for the long delay in reviewing. We have had many fires to fight
> with
> > the recent Chrome milestone. I am now actively re-leaning some of the bugs,
> and
> > experimenting with your CL and others.
> 
> Yeah, I know that. Heather told me that you are overloaded because of a very
> important Chrome milestone(but I don't know the exact day, :) ). I hope you
can
> take a look when you have time.

In fact, this CL's logic is basically the same as that in
SkTextToPathIter::SkTextToPathIter in SkPaint.cpp, who is called by
SkDraw::drawText_asPaths. Unfortunately,  currently Skdraw::drawPosText_asPaths
doesn't follow the logic in SkDraw::drawText_asPaths, and show rendering errors.