chore: update deprecated API usage in hello-python-ai examples

[jsonbailey]

Summary

Audits and updates the Python AI SDK example apps to stop using deprecated API surfaces. Doc guides reference these examples, so they need to reflect current SDK usage.

Changes

• examples/openai_example.py — Replaced deprecated tracker.track_openai_metrics(...) with tracker.track_metrics_of(..., get_ai_metrics_from_response) (imported from ldai_openai). Added flush() before close().
• examples/chat_judge_example.py — Replaced deprecated aiclient.create_chat(...) with aiclient.create_model(...). Fixed import from ldai to ldai.client. Added flush() before close().
• examples/chat_observability_example.py — Same create_chat → create_model swap and ldai → ldai.client import fix. Added flush() before close().
• examples/direct_judge_example.py — Fixed import from ldai to ldai.client. Added flush() before close().
• examples/bedrock_example.py, gemini_example.py, langchain_example.py, langgraph_agent_example.py, langgraph_multi_agent_example.py — Added ldclient.get().flush() before ldclient.get().close() so pending events are sent before shutdown.

No functional behavior was changed — only API calls were updated to use current methods.

Test plan

• Run poetry run chat-judge-example and confirm it executes end-to-end
• Run poetry run openai-example and confirm metrics are tracked via the new extractor path
• Spot-check the remaining examples still run without import errors

🤖 Generated with Claude Code