libvpx: Pull from upstream

libvpx: Pull from upstream

This is a partial checkin. It's too large for gcl.

This overwrites the m31 merge. The log that follows is from the previous
checkin and includes upstream fixes for:
https://codereview.chromium.org/25512006
https://codereview.chromium.org/23922008
and does not modify build/make/ads2gas_apple.pl to preserve:
https://codereview.chromium.org/24016003

Current HEAD: 25a2740abb59c90f481f170b10397728a54e3550

git log from upstream:
25a2740 Merge "Reducing the number of foreach_transformed_block() calls."
47b6030 Reducing the number of foreach_transformed_block() calls.
d515716 Merge "mb_lpf_horizontal_edge AVX2 optimization"
d03b3cb Merge "Fix x_offset_q4/y_offset_q4 calculation"
54f9205 mb_lpf_horizontal_edge AVX2 optimization
b26ce8b Updates to 1-pass:
a2a5c6f Merge "Enable all-zero coeff block index for sub8x8 blk"
ca39a00 Merge "Reducing the number of recursive calls."
6761872 Replacing (SWITCHABLE_FILTERS + 1) with SWITCHABLE_FILTER_CONTEXTS.
8c8381d Enable all-zero coeff block index for sub8x8 blk
2901bf2 Reducing the number of recursive calls.
54c2854 vp9/decode: align tile worker data allocation
2a67a34 Merge "vp9_decodframe.c: use vpx_memset instead of cast"
b18ea70 Merge "idct_blk_mmx.c: use vpx_memset instead of cast"
272d76e Merge "decodframe.c: use vpx_memset instead of cast"
d03d960 Merge "threading.c: use vpx_memset instead of cast"
15a24c6 Merge "idct_blk.c: use vpx_memset instead of cast"
9ed2d0a Fix x_offset_q4/y_offset_q4 calculation
ce053e7 Merge "vp9: add multi-threaded tile decoder"
3b47e05 Merge "vp9/decode: add get_tile()"
<...>

See full log with git log c701eeb..25a2740

R=tomfinegan@chromium.org
TBR=tomfinegan@chromium.org

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=232229

[Tom Finegan, 2013-10-31 22:54:40]

lgtm/rubber stamp

[wwcv, 2013-10-31 22:55:22]

Committed patchset #1 manually as r232229 (presubmit successful).