direct text2video / cogvideo?

[sbaier1]

new mechanism for cogvideo or other general text2video mechanism or perhaps mechanisms that can animate frames generated by other mechanisms in some way

API changes:

• probably need to make mechanisms return a list of frames so it can be 1..n frames generated instead of just one and iterating as it is now
• mechanisms need to become slightly more stateful, an API to pass frames from the previous scene is probably in order