Remove isIRect "optimization" & defer computation of device Rect in AA rect rendering

This doesn't seem to make much of a difference either way performance-wise (at least on Windows with a repeat count of 100). These 4 tests survived a pruning of anything <= .5ms and <= 3% and appeared in two consecutive tests (old is on left; new is on right):

Test1

     blur_10_inner_low_quality GPU c          4.52       5.30      -0.78   -17.3% 
blur_100.50_normal_low_quality GPU c          8.11       9.05      -0.94   -11.6% 

                  arbroundrect GPU c         47.42      45.86      +1.56    +3.3%
          giantdashline_hori_2 GPU c         42.43      40.25      +2.18    +5.1%

Test2:

      blur_10_outer_low_quality GPU c          4.21       4.84      -0.63   -15.0% 
 blur_100.50_normal_low_quality GPU c          7.96       9.05      -1.09   -13.7% 

                   arbroundrect GPU c         48.05      46.49      +1.56    +3.2%
           giantdashline_hori_2 GPU c         43.37      41.34      +2.03    +4.7%

[brian1, 2013-05-10 02:56:26]

On 2013/05/09 20:13:11, robertphillips wrote:
Lgtm

[commit-bot: I haz the power, 2013-05-10 10:14:56]

CQ is trying da patch. Follow status at
https://skia-tree-status.appspot.com/cq/robertphillips@google.com/14890021/1

[commit-bot: I haz the power, 2013-05-10 10:22:37]

Retried try job too often on Build-Ubuntu12-GCC-x86_64-Release-Trybot for
step(s) BuildBench, BuildGM, BuildMost, BuildSkiaBaseLibs, BuildTests,
BuildTools
http://108.170.217.252:10117/buildstatus?builder=Build-Ubuntu12-GCC-x86_64-Re...

[robertphillips, 2013-05-10 11:06:57]

committed as r9087