feat: Update cohere embedding models

[vblagoje]

Related Issues

• No related issue.
• Cohere announced a new multilingual embedding model today, so we are adding it to Haystack. The new model should perform better than any other open-source embedder while projecting in 768 dimensions - excellent!
• Enabling the use of finance-sentiment embedding model. This is not a baseline model but is listed and available as a "Cohere finetuned model"

Proposed Changes:

Adds a new model to cohere embedder. Use it by simply specifying it:

retriever = EmbeddingRetriever(
    document_store=document_store,
    embedding_model="multilingual-22-12", 
    batch_size = 16,
    api_key=my_cohere_api_key,
)

How did you test it?

I ran a Collab notebook to try it out. It worked as expected.

Notes for the reviewer

Try out the notebook yourself.

Checklist

• I have read the contributors guidelines and the code of conduct
• I have updated the related issue with new insights and changes
• I added tests that demonstrate the correct behavior of the change
• I've used the conventional commit convention for my PR title
• I documented my code
• I ran pre-commit hooks and fixed any issue

[masci]

LGTM, the change seems safe enough I propose we ship this in 1.12