Replacing (raster_block >> tx_size) with (block >> (tx_size << 1)).

The new expression is much more logical than previous one. Surprisingly
both expressions give exactly the same set of dependent values
-- have_top, have_left, have_right -- in vp9_predict_intra_block.

Change-Id: I63eb1b592b8c37883b3a0dbb1f3daa271e446109

vp9/decoder/vp9_decodframe.c

diff --git a/vp9/decoder/vp9_decodframe.c b/vp9/decoder/vp9_decodframe.c
index 4746a3a..86fba05 100644
--- a/vp9/decoder/vp9_decodframe.c
+++ b/vp9/decoder/vp9_decodframe.c
@@ -322,7 +322,7 @@
   if (xd->mb_to_right_edge < 0 || xd->mb_to_bottom_edge < 0)
     extend_for_intra(xd, plane_bsize, plane, block, tx_size);
 
-  vp9_predict_intra_block(xd, raster_block >> tx_size,
+  vp9_predict_intra_block(xd, block >> (tx_size << 1),
                           b_width_log2(plane_bsize), tx_size, mode,
                           dst, pd->dst.stride, dst, pd->dst.stride);