feat: Add support for Ollama

[Hari-Jeon]

This PR adds support for Ollama integration, allowing users to connect NotebookLlama with Ollama's model-serving framework while maintaining full backward compatibility.

Changes

• New: Ollama configuration via OLLAMA_BASE_URL env variable
• New: Optional LLM specification via OPENAI_API_MODEL and OLLAMA_MODEL env variables
• Added: Logger warnings when components like the PodcastGenerator or other variables are not initialized
• Updated: All LLM client instantiation to optionally use Ollama
• Updated: Documentation in README.md and .env.example to include Ollama setup instructions

Configuration

• Default: NotebookLlama works exactly as before
• Ollama: Set OLLAMA_BASE_URL in your .env file to use Ollama models.
• LLM: Set OPENAI_API_MODEL or OLLAMA_MODEL in your .env file to change the default LLM

Backward Compatibility

• 100% backward compatible — existing code and setups require no changes
• Optional feature - you can opt into Ollama LLMs and change the default OpenAI "gpt-4.1" model.

This addresses the Ollama integration feature discussed in #10

[AstraBert]

I'm not 100% sure this is a valuable addition: if we want to go local, we should be going all-in and replace all the remote services, otherwise I feel like this just complicates things

[AstraBert]

Also all the CI checks are failing, we would need to make them pass before considering merging this change

[Hari-Jeon]

Thanks for the feedback. I'll work on it some more. Until then, I'll close the pr and open it again later with an upgraded version.