Add Ollama provider support for v0.2.2

[HuRuilizhen]

Summary

This PR is the first v0.2.2 slice for strengthening the LLM concept-extraction path.

The main goal is to add Ollama as a supported local LLM provider while preserving the current llm extractor contract, the synchronous public pipeline API, and the existing downstream ConceptMention shape.

Included

• adds Ollama provider support

  • extends the supported llm provider set to include ollama
  • adds a default local Ollama base URL of http://localhost:11434/v1
  • keeps Ollama on the current OpenAI-compatible chat-completions path
  • treats API keys as optional for the local Ollama provider

• adds focused provider coverage for Ollama

  • verifies the default Ollama base URL
  • verifies Ollama requests do not require an authorization header when no key is configured
  • preserves the existing provider regression coverage for other LLM backends

• updates the existing LLM extraction example

  • extends examples/llm_extraction.py to show how to run against a local Ollama model
  • keeps one shared LLM example rather than introducing a provider-specific duplicate script

Validation

Repository checks:

• ruff check .
• pyright
• pytest -q

Targeted checks:

• pytest -q tests/test_llm_provider.py

Notes

• this PR does not yet introduce the broader output-contract capability adaptation planned for 0.2.2; that remains the next slice
• local-only planning documents under docs/ remain intentionally untracked and are not part of this branch