[AI] Text to video pipeline support

[stronk-dev]

This PR adds support for a text-to-video pipeline

Rebased my ancient text-to-video pipeline and some stuff might've broken, so needs a thorough review.

Also let me know if the refactor from ImageToVideoResult -> VideoResult is desired or not. I can imagine this might cause merge conflicts with other PR's so might be better to separate it out and add a TextToVideoResult for now.

Dependent on https://github.com/livepeer/ai-worker/pull/187