Ollama Integration Fix

In escargot/language_models/ollama.py, the get_embedding() function on line 139 should use the embedding model ID and extract the embedding using the "embedding" key from the response. The fixed function should be:

def get_embedding(self, text_to_embed):
response = ollama.embed(model=self.embedding_id, input=text_to_embed)
return response["embeddings"][0]

Additionally, escargot needs to be installed from the git repo for ollama integration to work; the PyPI version is likely not up to date.