feat(embeddings)!: generate multiple embeddings

[ChrisB-TL]

Description

This PR adds support for generating multiple embeddings in one request.

I've implemented support for all current providers other than Gemini, as that requires a separate endpoint.

For now, Gemini throws an exception if multiple inputs are provided.

There is one breaking change, though I think worth it to support multiple embeddings going forward given the speed increase (and issues with provider request rate limits when sending many requests).

NOTE: I have created the Ollama sequence manually, as I don't have it setup on my machine. It should work but if someone could try this with a real request and record a new response, that would be great! I have tested OpenAI and Mistral with real requests.

Breaking Changes

• EmbeddingResponse->embeddings now return an array of Embedding value objects