Add zero code Ollama example

[ajimenez1503]

Summary

Adds a new zero-code local-model example for Ollama using the OpenAI-compatible API path, and documents it in the examples guide.

This PR addresses the local-dev use case where users do not have an OPENAI_API_KEY and want to run agentevals with a local model.

Fix #90

Running example

python3 run.py

OTLP endpoint: http://localhost:4318
Local model endpoint: http://localhost:11434/v1
Local model: llama3.2:3b

[1/3] User: Hi! Can you help me?
     Agent: You rolled a 2 on a 20-sided die. Would you like to roll again?

[2/3] User: Roll a 20-sided die for me
     Agent: You rolled a 10 on a 20-sided die. Would you like to roll again?

[3/3] User: Is the number you rolled prime?
     Agent: The number 10 is not a prime number. Would you like to roll again or ask another question?

All traces and logs flushed to OTLP receiver.

Looks good in agentEval UI

[antonjim-te]

/review @krisztianfekete

[krisztianfekete]

Wow, this model is not too great. Maybe we could even drop tool usage altogether from this example, or use a larger model. Anyway, merging this now and we can simplify it later as the actual example does work. Thank you for the PR!