Makes second filters for 45-degree directions horizontal

This reduces the deringing worst-case number of "latency" lines from 10 to 8.

ntt-short1 results:
         MEDIUM (%) HIGH (%)
    PSNR -0.063681 -0.027661
 PSNRHVS -0.063628 -0.007841
    SSIM -0.007828 -0.048415
FASTSSIM  0.139420 -0.183910

Change-Id: I006544ab748b76c0334e406300b3eda20e125846

vp10/common/od_dering.c

diff --git a/vp10/common/od_dering.c b/vp10/common/od_dering.c
index 5b9edc3..be6d48e 100644
--- a/vp10/common/od_dering.c
+++ b/vp10/common/od_dering.c
@@ -164,7 +164,7 @@
   int i;
   int j;
   int offset;
-  if (dir <= 4) offset = OD_FILT_BSTRIDE;
+  if (dir > 0 && dir < 4) offset = OD_FILT_BSTRIDE;
   else offset = 1;
   for (i = 0; i < 1 << ln; i++) {
     for (j = 0; j < 1 << ln; j++) {