Add support or guidance for multimodal ML workflows in MLOps Stacks

[davidzuma]

I would like to request support (or at least documentation and examples) for multimodal machine learning workflows within the Databricks MLOps Stacks.

Currently, the stack appears well-suited for traditional single-modality use cases, but as ML systems evolve, many modern applications involve multiple models orchestrated into a single model and for that reason I'd suggest suggest one (or more) of the following:

• Add out-of-the-box templates that demonstrate managing and deploying multimodal workflows.

• Provide official guidance on how to extend the existing stack for multimodal setups (e.g., handling multiple model artifacts, data modalities, and evaluation metrics).

• Clarify whether the current stack design already allows such workflows through configuration, and if so, provide an example.

Thank you for your time and for continuing to evolve this excellent project.