Add gemini-embedding-001 as embedding model for Gemini

[jperezdealgaba]

🚀 Describe the new functionality needed

The Gemini inference provider should be able to use the gemini-embedding-001 model instead of text-embedding-004 :

llama-stack/llama_stack/providers/remote/inference/gemini/gemini.py

Line 17 in 4b9ebbf

"text-embedding-004": {"embedding_dimension": 768, "context_length": 2048},

Google is deprecating this model and soon no embedding will be able to be performed

💡 Why is this needed? What if we don't build it?

Google is deprecating old embedding models: Link and soon the only allowed model for Gemini won't be able to be used

Other thoughts

No response