Add Ollama option to choice of LLMs

[rchan26]

As discussed in #97. I have added an option to use Ollama as the LLM. By default the model is a quantised SmolLM-1.7B-Instruct.

For example, if you're running an Ollama server from http://127.0.0.1:8000, you can run:

python s2s_pipeline.py \
    --local \
    --device mps \
    --llm ollama \
    --ollama_api_endpoint http://127.0.0.1:8000 

By default the ollama_api_endpoint argument is http://localhost:11434 which is the default host when spinning up an Ollama endpoint.

Note that you must have downloaded the model for Ollama prior to running, so first run

ollama pull hf.co/HuggingFaceTB/SmolLM2-1.7B-Instruct-GGUF

before starting an endpoint with

ollama serve

Also updated a few help comments.