Replace _mm_cvtps_epi32(x) with _mm_cvttps_epi32(_mm_add_ps(0.5f), x).

Replace _mm_cvtps_epi32(x) with _mm_cvttps_epi32(_mm_add_ps(0.5f), x).

We don't have control over which way _mm_cvtps_epi32 rounds.

  - This makes the SSE SkPMFloat rounding consistent with _neon and _none.
  - Sk4f::cast<Sk4i>() is closer to (int)float's behavior.  (Correct when >=0).

Add tests that would fail at head.

BUG=skia:

Committed: https://skia.googlesource.com/skia/+/92d04da38f03dfabd8cd9a7244588a49be9a2f41

[mtklein_C, 2015-03-23 18:44:06]

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

[reed1, 2015-03-23 18:49:06]

lgtm

[mtklein, 2015-03-23 18:56:15]

The CQ bit was checked by mtklein@google.com

[commit-bot: I haz the power, 2015-03-23 18:56:27]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1029163002/40001

[commit-bot: I haz the power, 2015-03-23 19:01:50]

Committed patchset #3 (id:40001) as
https://skia.googlesource.com/skia/+/92d04da38f03dfabd8cd9a7244588a49be9a2f41