fix(langchain-openai): do not subtract priority token counts

[hassiebp]

Only subtract known token detail subcategories in _parse_usage_model to fix incorrect input/output totals caused by priority-tier metadata (fixes #11896).

---

 

---

Important

Fix issue #11896 by updating _parse_usage_model to skip subtraction for 'priority' token details and adding tests in test_langchain_usage.py.

• Behavior:
  • Modify _parse_usage_model in CallbackHandler.py to skip subtraction for token details with keys starting with "priority".
  • Introduce _should_subtract_token_detail() to determine if a token detail should be subtracted.
• Tests:
  • Add test_langchain_usage.py with tests test_parse_usage_model_skips_priority_subtraction and test_parse_usage_model_subtracts_known_details to verify correct behavior of _parse_usage_model.

^{This description was created by for 04fca34. You can customize this summary. It will automatically update as commits are pushed.}

[cursor]

Cursor Agent can help with this pull request. Just @cursor in comments and I'll start working on changes in this branch.
_{Learn more about Cursor Agents}