vp10: fix 4:2:2 chroma MVs for 8x4/4x4 blocks.

In vp9, the bottom MV would be the average of the topright and
bottomleft luma MV (instead of the bottomleft/bottomright luma MV).

See issue 993.

Change-Id: Ic91c0b195950e7b32fc26c84c04788a09321e391

vp10/common/reconinter.c

@@ -143,11 +143,11 @@ static void build_inter_predictors_for_planes(MACROBLOCKD *xd, BLOCK_SIZE bsize,
     const int bh = 4 * num_4x4_h;
 
     if (xd->mi[0]->mbmi.sb_type < BLOCK_8X8) {
-      int i = 0, x, y;
+      int x, y;
       assert(bsize == BLOCK_8X8);
       for (y = 0; y < num_4x4_h; ++y)
         for (x = 0; x < num_4x4_w; ++x)
-           build_inter_predictors(xd, plane, i++, bw, bh,
+           build_inter_predictors(xd, plane, y * 2 + x, bw, bh,
                                   4 * x, 4 * y, 4, 4, mi_x, mi_y);
     } else {
       build_inter_predictors(xd, plane, 0, bw, bh,

vp10/decoder/decodeframe.c

@@ -707,10 +707,10 @@ static void dec_build_inter_predictors_sb(VP10Decoder *const pbi,
       const int is_scaled = vp10_is_scaled(sf);
 
       if (sb_type < BLOCK_8X8) {
-        int i = 0, x, y;
+        int x, y;
         for (y = 0; y < num_4x4_h; ++y) {
           for (x = 0; x < num_4x4_w; ++x) {
-            const MV mv = average_split_mvs(pd, mi, ref, i++);
+            const MV mv = average_split_mvs(pd, mi, ref, y * 2 + x);
             dec_build_inter_predictors(pbi, xd, plane, n4w_x4, n4h_x4,
                                        4 * x, 4 * y, 4, 4, mi_x, mi_y, kernel,
                                        sf, pre_buf, dst_buf, &mv,