fix: strip thinking tags before all JSON parse strategies

[haosenwang1018]

Closes #159

When using reasoning models (qwen2.5-think, deepseek-r1, etc.), the <think>...</think> tags were only stripped in _extract_all_json_candidates(), meaning the regex fallback strategy (_extract_fields_with_regex) still operated on the raw response including thinking content. This could cause it to extract content from the thinking section rather than the actual analysis.

This fix moves the think-tag stripping to the top of _robust_json_parse() so all downstream strategies work with clean model output.

[LarFii]

Thanks for the fix.

I do see one potential side effect to consider:

• Potential data loss from global tag stripping
  The current implementation removes <think>...</think> / <thinking>...</thinking> across the entire response before all parsing strategies.
  If those tags appear as legitimate literal content inside the actual payload (e.g., in detailed_description), that content would be removed unintentionally.

Also a minor maintainability point:

• Duplicated cleanup logic
  The same think-tag stripping still exists in _extract_all_json_candidates(), so cleanup now happens in two places. It may be better to centralize this in one function to avoid drift.

Suggested refinement

Instead of global removal, strip only leading reasoning blocks (prefix-only), which still fixes the fallback issue while avoiding accidental mutation of valid body content.