Support for image embeddings in auto-embeddings mode

[nepridumalnik]

Proposal:

Description:

Currently, auto-embeddings in ManticoreSearch only works with text models from HuggingFace. There is no way to use it for image-to-image or text-to-image search.

Proposal:

1. Extend auto-embeddings to support image embeddings.
2. Allow CLIP-based models (OpenAI CLIP, OpenCLIP, TinyCLIP) to generate embeddings for both text and images.
3. Enable multimodal search scenarios:

• search by image → find visually similar images
• search by text → find matching images

Use case:

A user has a database of product images (e.g., clothing from marketplaces). They want to:

• Search for "gray coat" using a text query.
• Upload an image of a coat and find visually similar coats.

Checklist:

^{To be completed by the assignee. Check off tasks that have been completed or are not applicable.}

• [ ] Implementation completed
• [ ] Tests developed
• [ ] Documentation updated
• [ ] Documentation reviewed
• [x] OpenAPI YAML updated and issue created to rebuild clients