Add support for Image/Multimodal Model

[atarora]

Considering Qdrant would support "any" kind of embedding, we should have a way to process image/multimodal embeddings.

Is there an existing way to do it through Fastembed?

[NirantK]

Can consider adding CLIP support #70

[NirantK]

Is there anything other than images or non-CLIP models which we're interested in?

[NirantK]

Closing this as duplicate of #70 for now