Fix provider lazy loading to prevent API key errors for local models

[aksg87]

Description

Fixes #110 - Local Ollama models were incorrectly requiring API keys because provider patterns weren't registered before model resolution.

Root cause: PR #97's lazy loading meant providers weren't imported when registry.resolve() was called, causing fallback to cloud providers.

Bug fix

Solution

• Load providers before pattern matching in factory.create_model()
• Skip loading when provider is explicitly specified (optimization)
• Remove redundant eager imports that defeated lazy loading

Changes

1. Added load_builtins_once() to ensure providers are registered before resolution
2. Optimized to only load when pattern matching is needed
3. Removed duplicate module-level imports in providers/__init__.py
4. Added regression test for canonical Ollama models

How Has This Been Tested?

pytest tests/factory_test.py -v
# 18 passed

# New test covers: llama3, gemma2:2b, mistral:7b, qwen3:0.6b
pytest tests/factory_test.py::FactoryTest::test_ollama_models_select_without_api_keys -v

Checklist:

• I have read and acknowledged Google's Open Source Code of conduct.
• I have read the Contributing page, and I either signed the Google Individual CLA or am covered by my company's Corporate CLA.
• I have discussed my proposed solution with code owners in the linked issue(s) and we have agreed upon the general approach.
• I have made any needed documentation changes, or noted in the linked issue(s) that documentation elsewhere needs updating.
• I have added tests, or I have ensured existing tests cover the changes
• I have followed Google's Python Style Guide and ran pylint over the affected code.