Add additional source types to xfer mode test.

Add additional source types to xfer mode test.

Committed: http://code.google.com/p/skia/source/detail?r=1705

[Steve VanDeBogart, 2011-06-13 22:32:50]

The PDF code doesn't work properly for Clear, Src, SrcIn, SrcOut, DstIn, DstOut.
 Because the drew false conclusion about what they meant from the xfermode gm
test.  This expands the test to demonstrate the functionality better.  If you
are ok with this change, I'll regenerating baselines.  The result of this change
(xfermodes_8888.png) is in my web directory.

[reed1, 2011-06-14 13:48:47]

1. Can you document why we don't use all of the modes each time (i.e. what the
mask bits represent and why)?

2. When you mul or divide, often you can avoid using SkScalarMul and
SkScalarDiv. e.g.

   SkScalarMul(H, SK_ScalarHalf) == H / 2

   SkScalarDiv(W * 37, SkIntToScalar(60)) == W * 37 / 60

3. Skia wants 4-space indents

[Steve VanDeBogart, 2011-06-24 02:03:44]

On 2011/06/14 13:48:47, reed1 wrote:
> 1. Can you document why we don't use all of the modes each time (i.e. what the
> mask bits represent and why)?

Done.

> 2. When you mul or divide, often you can avoid using SkScalarMul and
> SkScalarDiv. e.g.
> 
>    SkScalarMul(H, SK_ScalarHalf) == H / 2
> 
>    SkScalarDiv(W * 37, SkIntToScalar(60)) == W * 37 / 60

Done.
 
> 3. Skia wants 4-space indents

I thought I had vim doing that automagically - fixed.

This now includes all the rebaselines except macbook gpu (macbook/macpro sw are
the same).

[reed1, 2011-06-24 12:26:11]

LGTM