aacc + bbdd

aacc + bbdd

SkMatrix::mapPts() using aacc/bbdd was always worse than using badc():
  - On Intel, it was faster than exisiting swizzle, but badc() is 10% faster still (one pshufd instead of two).
  - On ARM, existing swizzle < badc() < aacc()+bbdd(), even though aacc() then bbdd() is really a single vtrn instruction.

I will revert SkMatrix.cpp before submitting.  Just thought you might like to look.

Will think more and try to gear up Instruments on ARM.

BUG=skia:

Committed: https://skia.googlesource.com/skia/+/470fa6f74b8ed08d85af463d5b3ef6f19ed6522a

[mtklein_C, 2015-03-24 01:16:43]

mtklein@chromium.org changed reviewers:
+ reed@google.com

[reed1, 2015-03-24 12:29:38]

lgtm

[mtklein, 2015-03-24 13:11:23]

The CQ bit was checked by mtklein@google.com

[mtklein, 2015-03-24 13:11:24]

The patchset sent to the CQ was uploaded after l-g-t-m from reed@google.com
Link to the patchset: https://codereview.chromium.org/1012573003/#ps20001
(title: "revert SkMatrix.cpp")

[commit-bot: I haz the power, 2015-03-24 13:11:32]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1012573003/20001

[commit-bot: I haz the power, 2015-03-24 13:16:34]

Committed patchset #2 (id:20001) as
https://skia.googlesource.com/skia/+/470fa6f74b8ed08d85af463d5b3ef6f19ed6522a