vp9: Fix a bug where best_sse_mv is not set properly.

In VP9 internal denoiser, motion magnitude is computed from
best_sse_mv, which should be set to 0 at the begining. This bug may
cause visual aritifact in denoiser. Also, delete two improper comments.

Change-Id: I8710d2acba23320bc85cf72af17d65245c19438b

vp9/encoder/vp9_denoiser.c

diff --git a/vp9/encoder/vp9_denoiser.c b/vp9/encoder/vp9_denoiser.c
index 7f5265a..42d456e 100644
--- a/vp9/encoder/vp9_denoiser.c
+++ b/vp9/encoder/vp9_denoiser.c
@@ -21,12 +21,6 @@
 #include "vp9/encoder/vp9_denoiser.h"
 #include "vp9/encoder/vp9_encoder.h"
 
-/* The VP9 denoiser is similar to that of the VP8 denoiser. While
- * choosing the motion vectors / reference frames, the denoiser is run, and if
- * it did not modify the signal to much, the denoised block is copied to the
- * signal.
- */
-
 #ifdef OUTPUT_YUV_DENOISED
 static void make_grayscale(YV12_BUFFER_CONFIG *yuv);
 #endif
@@ -500,12 +494,12 @@
   ctx->zeromv_sse = UINT_MAX;
   ctx->newmv_sse = UINT_MAX;
   ctx->zeromv_lastref_sse = UINT_MAX;
+  ctx->best_sse_mv.as_int = 0;
 }
 
 void vp9_denoiser_update_frame_stats(MODE_INFO *mi, unsigned int sse,
                                      PREDICTION_MODE mode,
                                      PICK_MODE_CONTEXT *ctx) {
-  // TODO(tkopp): Use both MVs if possible
   if (mi->mv[0].as_int == 0 && sse < ctx->zeromv_sse) {
     ctx->zeromv_sse = sse;
     ctx->best_zeromv_reference_frame = mi->ref_frame[0];