Support for `multimodalembedding` image embeddings

[ajroetker]

This library doesn't support the Vertex Google Model multimodalembedding functionality of creating embeddings from images and fails silently when you do pass image data to the model.

https://github.com/googleapis/go-genai/blob/main/transformer.go#L157 Should support the image data in the inlineData field as well as text instead of just text. Probably should support the gcsUri too since the model supports that.

Implementing manually works https://cloud.google.com/vertex-ai/generative-ai/docs/embeddings/get-multimodal-embeddings.