RISC-V harness assumes V

But The current RISC-V hardness is not usable in FFmpeg as it stands because it assumes that V is present. FFmpeg has a few non-V optimisations, and the current primary FATE runner does not support V (only RVA20 + Zba + Zbb).

In general, I am biased but I much prefer the FFmpeg RISC-V harness over dav1d's.