AVX2 YUV alpha blender and improved unittests

include/libyuv/planar_functions.h

Index: include/libyuv/planar_functions.h
diff --git a/include/libyuv/planar_functions.h b/include/libyuv/planar_functions.h
index 6d5dd082f9d09d9bc0860583c75c7f6227c39c07..9d30225d4cbb6d5bdef9dbbf2be9c15d603bc962 100644
--- a/include/libyuv/planar_functions.h
+++ b/include/libyuv/planar_functions.h
@@ -302,6 +302,7 @@ LIBYUV_API
 ARGBBlendRow GetARGBBlend();
 
 // Alpha Blend ARGB images and store to destination.
+// Source is pre-multiplied by alpha using ARGBAttenuate.
 // Alpha of destination is set to 255.
 LIBYUV_API
 int ARGBBlend(const uint8* src_argb0, int src_stride_argb0,
@@ -309,6 +310,31 @@ int ARGBBlend(const uint8* src_argb0, int src_stride_argb0,
               uint8* dst_argb, int dst_stride_argb,
               int width, int height);
 
+// Alpha Blend plane and store to destination.
+// Source is not pre-multiplied by alpha.
+LIBYUV_API
+int BlendPlane(const uint8* src_y0, int src_stride_y0,
+               const uint8* src_y1, int src_stride_y1,
+               const uint8* alpha, int alpha_stride,
+               uint8* dst_y, int dst_stride_y,
+               int width, int height);
+
+// Alpha Blend YUV images and store to destination.
+// Source is not pre-multiplied by alpha.
+// Alpha is full width x height and subsampled to half size to apply to UV.
+LIBYUV_API
+int I420Blend(const uint8* src_y0, int src_stride_y0,
+              const uint8* src_u0, int src_stride_u0,
+              const uint8* src_v0, int src_stride_v0,
+              const uint8* src_y1, int src_stride_y1,
+              const uint8* src_u1, int src_stride_u1,
+              const uint8* src_v1, int src_stride_v1,
+              const uint8* alpha, int alpha_stride,
+              uint8* dst_y, int dst_stride_y,
+              uint8* dst_u, int dst_stride_u,
+              uint8* dst_v, int dst_stride_v,
+              int width, int height);
+
 // Multiply ARGB image by ARGB image. Shifted down by 8. Saturates to 255.
 LIBYUV_API
 int ARGBMultiply(const uint8* src_argb0, int src_stride_argb0,