3-15% speedup to HardLight / Overlay xfermodes.

src/opts/Sk4px_SSE2.h

Index: src/opts/Sk4px_SSE2.h
diff --git a/src/opts/Sk4px_SSE2.h b/src/opts/Sk4px_SSE2.h
index 74ccffc277227432e5266e750f8aa48de34c8c42..3809c5e47b5a955dfe6eeb794722c6950d8bb010 100644
--- a/src/opts/Sk4px_SSE2.h
+++ b/src/opts/Sk4px_SSE2.h
@@ -31,6 +31,11 @@ inline Sk4px::Wide Sk4px::widenHi() const {
                  _mm_unpackhi_epi8(_mm_setzero_si128(), this->fVec));
 }
 
+inline Sk4px::Wide Sk4px::widenLoHi() const {
+    return Sk16h(_mm_unpacklo_epi8(this->fVec, this->fVec),
+                 _mm_unpackhi_epi8(this->fVec, this->fVec));
+}
+
 inline Sk4px::Wide Sk4px::mulWiden(const Sk16b& other) const {
     return this->widenLo() * Sk4px(other).widenLo();
 }