dav1d 0.2.1 'Antelope', the fast and small AV1 decoder

This is the second release of dav1d, the fast and small AV1 decoder,
codename 'Antelope'.

It still supports all AV1 features and all bitdepths, from 8 to 12bits,
and decodes more samples than before.

While still being the fastest decoder on AVX-2 CPUs, dav1d is now also
faster on ARM, ARM64 and SSSE3 CPUs in normal loads.
0.2.x is around 2x/2.5x faster on SSSE3 and achieves 1080p at 80fps
on an iPhone XS

0.2.1 is a small bugfix release above 0.2.0 to fix build issues, and
increase a bit the speed of SSSE3