Inference Text/Image to video - TiledVAE feature

[Domica]

Proposal

As proposed through Discord feature addition, new Inference option for Text/Image to video screen for using TiledVAE to avoid OOMs on low to mid VRAM GPUs.

Would be appreciated for newbies to easier use the appliation rather than entering the Comfyui flow/node view.

Have some AI code ideas (as only know superficialy coding).

Steps (assumed):

1. New XAML UI Card
   Path: StabilityMatrix.Avalonia/Controls/Inference/TiledVAEEncodeCard.axaml

2. Code for the card
   Path: StabilityMatrix.Avalonia/Controls/Inference/TiledVAEEncodeCard.axaml.cs

3. ViewModel for the card (with bindings & validation)
   Path: StabilityMatrix.Avalonia/ViewModels/Inference/TiledVAEEncodeCardViewModel.cs

4. Inference Module (core)
   Path: StabilityMatrix.Avalonia/ViewModels/Inference/Modules/TiledVAEEncodeModule.cs

5. Typed Node Definition for ComfyUI
   Add to existing file: StabilityMatrix.Core/Models/Api/Comfy/Nodes/ComfyNodeBuilder.cs

6. Model register

7. JSON Serialization Support

8. Avalonia Style Include (if needed for discovery)

Would be appreaciated by contributors to assess the idea for implementation.

Thanks

[Domica]

Forked Dev version, will see if can try adding the feature (fingers crossed)