[AI] Add Support for New Modalities / Content Types

[jeffhandley]

Overall, we need to:

• Get familiar with existing SDK APIs for new modalities and content types
  • Determine if there's enough coverage across providers to start to generalize
  • Consider using Deep Research to conduct this exercise itself
• Implement new content type abstractions, or prove existing abstractions suffice
• End-to-end scenario validation
  • End-to-end scenarios for bouncing between modalities
  • End-to-end scenarios around document/content creation, product support, etc.

Backlog

• Text to Video
• Image to Video
• Video to Video
• Image to Text