Optimize vp9_tree_probs_from_distribution

The previous implementation visited each node in the tree multiple times
because it used each symbol's encoding to revisit the branches taken and
increment its count. Instead, we can traverse the tree depth first and
calculate the probabilities and branch counts as we walk back up. The
complexity goes from somewhere between O(nlogn) and O(n^2) (depending on
how balanced the tree is) to O(n).

Only tested one clip (256kbps, CIF), saw 13% decoding perf improvement.

Note that this optimization should port trivially to VP8 as well. In VP8,
the decoder doesn't use this function, but it does routinely show up
on the profile for realtime encoding.

Change-Id: I4f2848e4f41dc9a7694f73f3e75034bce08d1b12

vp9/common/vp9_treecoder.h

@@ -47,12 +47,11 @@ void vp9_tokens_from_tree_offset(struct vp9_token_struct *, vp9_tree,
    taken for each node on the tree; this facilitiates decisions as to
    probability updates. */
 
-void vp9_tree_probs_from_distribution(int n,  /* n = size of alphabet */
-                                      vp9_token tok[ /* n */ ],
-                                      vp9_tree tree,
+void vp9_tree_probs_from_distribution(vp9_tree tree,
                                       vp9_prob probs[ /* n - 1 */ ],
                                       unsigned int branch_ct[ /* n - 1 */ ][2],
-                                      const unsigned int num_events[ /* n */ ]);
+                                      const unsigned int num_events[ /* n */ ],
+                                      unsigned int tok0_offset);
 
 static INLINE vp9_prob clip_prob(int p) {
   return (p > 255) ? 255u : (p < 1) ? 1u : p;