Add MSA optimized TransposeWx8_MSA and TransposeUVWx8_MSA functions

source/rotate.cc

Index: source/rotate.cc
diff --git a/source/rotate.cc b/source/rotate.cc
index 098fcf7c70e79484a944e000c1eb38a82ead1b61..bfd51421b3cbe299d36314e8fbc1b44dd33c6d89 100644
--- a/source/rotate.cc
+++ b/source/rotate.cc
@@ -62,6 +62,14 @@ void TransposePlane(const uint8* src,
     }
   }
 #endif
+#if defined(HAS_TRANSPOSEWX8_MSA)
+  if (TestCpuFlag(kCpuHasMSA)) {
+    TransposeWx8 = TransposeWx8_Any_MSA;
+    if (IS_ALIGNED(width, 16)) {
+      TransposeWx8 = TransposeWx8_MSA;
+    }
+  }
+#endif
 
   // Work across the source in 8x8 tiles
   while (i >= 8) {
@@ -232,6 +240,14 @@ void TransposeUV(const uint8* src,
     TransposeUVWx8 = TransposeUVWx8_DSPR2;
   }
 #endif
+#if defined(HAS_TRANSPOSEUVWX8_MSA)
+  if (TestCpuFlag(kCpuHasMSA)) {
+    TransposeUVWx8 = TransposeUVWx8_Any_MSA;
+    if (IS_ALIGNED(width, 8)) {
+      TransposeUVWx8 = TransposeUVWx8_MSA;
+    }
+  }
+#endif
 
   // Work through the source in 8x8 tiles.
   while (i >= 8) {