`StatelessVideoDecoder` implementations for `StatelessDecoder` can be factorized?

[Gnurou]

We have one implementation per codec, with many methods that are strictly identical. We should be able to factorize this by implementing the methods that differ as regular methods of StatelessDecoder, and having a single StatelessVideoDecoder impl block that calls these.

next_event would probably require another trait method on the decoder device.