Python: [Python][sample] sanitize JSON output in reflection pattern before Pydantic parsing (fixes #4718)

[hanhan761]

Summary

Add a JSON sanitization utility to the reflection pattern sample that handles
malformed model output (tab characters, markdown fences, truncated JSON) before
passing to Pydantic's \model_validate_json().

Changes

• \workflow_as_agent_reflection_pattern.py: added _sanitize_json_text()\ which:
  • Strips markdown code fences
  • Replaces tab characters with spaces
  • Attempts to fix truncated JSON by appending missing closing brackets/quotes

Issue

Fixes #4718

Verification

Cannot test locally (requires FoundryChatClient + gpt-4.1/gpt-5), but the
sanitization logic is isolated and straightforward.