Use Ollama client or LiteLLM for getting Ollama responses

[aponcedeleonch]

Right now we're calling directly Ollama API using httpx. Since the response is not standardized is causing problems when storing the responses from ollama in the DB. Potentially it can cause issues in more places of the pipeline. We should use instead:

• Ollama Python client
• Ollama LiteLLM

[lukehinds]

@lukehinds roadmap planning

[aponcedeleonch]

Closed via #241