[VP9][decode] the MD5 values are not the same as the reference data when using the ffmpeg contents

[keqiaox]

1.Testing environment: Libva: (master) d3b24b76dfde4daef9d463743971c7522a307092 Libva_intel_driver: (master) 4640d06a8ccb0a12e91bbd00cc3e2423d17ee243 Cmrt: (master)a4b0f2aefabfdbaf020f617deee40bc07b5992c9 intel-hybrid-driver(master)=0375a7287e556fedaf7628a85b143c1233a9ef84 ffmpeg: (master)=9ac61e73d0843ec4b83f4e3d47eded73234e406e libyami: (master)=8fe2149fa5ddba8a47fb43c0f371dc7681186e9c

2.Test command: $ ./yamidecode -i ./$filename -m -2 -f I420

3.Failed cases: 1---vp90-2-05-resize.ivf 2---vp90-2-10-show-existing-frame.webm

[ghost]

1.0.0 RC2 test update, remove odd resolution case, and then only one fail case: vp90-2-10-show-existing-frame.webm

[ghost]

1.0.1 RC1 release test update: fail case:

1. vp90-2-10-show-existing-frame.webm
2. vp90-2-16-intra-only.webm

[xuguangxin]

@wang-fei , this mean we have one new clip failed?

[xuguangxin]

@feiwan1 , please help update this.

[feiwan1]

@xuguangxin , based on libyami 1.2.0 RC2 test, there are still two case fail: vp90-2-10-show-existing-frame.webm vp90-2-16-intra-only.webm