Add test to exercise extreme blur sigmas

This test would've caught the pinterest blur crash (in the .1 & .01 sigma, cpu, rect-special-case cases).

Going forward I think we need to:

1) fix the underlying crash in the 9 patch code
2) update the blur code to pitch super small blurs
3) improve the non-rect-special-case cpu blur path to better match the ground truth (right now it can be 15 off)

[robertphillips, 2013-09-05 13:30:38]

https://codereview.chromium.org/23530039/diff/3001/src/effects/SkBlurMask.cpp
File src/effects/SkBlurMask.cpp (right):

https://codereview.chromium.org/23530039/diff/3001/src/effects/SkBlurMask.cpp...
src/effects/SkBlurMask.cpp:875: 
The 4x window still had some edge effects (when compared to the brute force 1D
convolution).

[humper, 2013-09-05 14:19:01]

On 2013/09/05 13:30:38, robertphillips wrote:
> https://codereview.chromium.org/23530039/diff/3001/src/effects/SkBlurMask.cpp
> File src/effects/SkBlurMask.cpp (right):
> 
>
https://codereview.chromium.org/23530039/diff/3001/src/effects/SkBlurMask.cpp...
> src/effects/SkBlurMask.cpp:875: 
> The 4x window still had some edge effects (when compared to the brute force 1D
> convolution).

[humper, 2013-09-05 14:19:17]

On 2013/09/05 14:19:01, humper wrote:
> On 2013/09/05 13:30:38, robertphillips wrote:
> >
https://codereview.chromium.org/23530039/diff/3001/src/effects/SkBlurMask.cpp
> > File src/effects/SkBlurMask.cpp (right):
> > 
> >
>
https://codereview.chromium.org/23530039/diff/3001/src/effects/SkBlurMask.cpp...
> > src/effects/SkBlurMask.cpp:875: 
> > The 4x window still had some edge effects (when compared to the brute force
1D
> > convolution).

lgtm

Nice catch on the ground truth error.

[humper, 2013-09-05 14:19:23]

https://codereview.chromium.org/23530039/diff/3001/src/effects/SkBlurMask.cpp
File src/effects/SkBlurMask.cpp (right):

https://codereview.chromium.org/23530039/diff/3001/src/effects/SkBlurMask.cpp...
src/effects/SkBlurMask.cpp:888: float gaussian = expf(-x*x / (2*variance));
WHOOPS, nice catch.

[bsalomon, 2013-09-05 14:23:59]

I defer to humper@

[robertphillips, 2013-09-05 16:39:21]

committed as r11099