Implement a NEON version of morphology. This is good for ~2.2X speedup on Tegra3.

src/opts/SkMorphology_opts_neon.h

Index: src/opts/SkMorphology_opts_neon.h
diff --git a/src/opts/SkMorphology_opts_SSE2.h b/src/opts/SkMorphology_opts_neon.h
similarity index 65%
copy from src/opts/SkMorphology_opts_SSE2.h
copy to src/opts/SkMorphology_opts_neon.h
index bd103e6eba9a6261f6f59ff7e61450acfd169373..0b962bdad61d4a71ed4ce6938a09a855e1c720cf 100644
--- a/src/opts/SkMorphology_opts_SSE2.h
+++ b/src/opts/SkMorphology_opts_neon.h
@@ -5,11 +5,11 @@
  * found in the LICENSE file.
  */
 
-void SkDilateX_SSE2(const SkPMColor* src, SkPMColor* dst, int radius,
+void SkDilateX_neon(const SkPMColor* src, SkPMColor* dst, int radius,
                     int width, int height, int srcStride, int dstStride);
-void SkDilateY_SSE2(const SkPMColor* src, SkPMColor* dst, int radius,
+void SkDilateY_neon(const SkPMColor* src, SkPMColor* dst, int radius,
                     int width, int height, int srcStride, int dstStride);
-void SkErodeX_SSE2(const SkPMColor* src, SkPMColor* dst, int radius,
+void SkErodeX_neon(const SkPMColor* src, SkPMColor* dst, int radius,
                    int width, int height, int srcStride, int dstStride);
-void SkErodeY_SSE2(const SkPMColor* src, SkPMColor* dst, int radius,
+void SkErodeY_neon(const SkPMColor* src, SkPMColor* dst, int radius,
                    int width, int height, int srcStride, int dstStride);