Support provider Ollama for automatic configuration.

[namin]

Ollama is already supported manually, as I describe in this discussion item: #379

Here, I make it possible to support Ollama in automatic configurations, e.g., as conveniently as:

export CONTROLFLOW_LLM_MODEL=ollama/qwen2.5

I updated the docs here:
https://controlflow.ai/guides/configure-llms#automatic-configuration
with a line for Ollama in supported providers for automatic configuration.

I also added a test in tests/llm/test_models.py, and accordingly added langchain-ollama as an optional tests dependency.

I systematically calqued all these changes by following the occurrences of the already supported provider groq in the repository.

Thanks!