fix: remove stream_usage from text completion

[Pouyanpi]

This pull request includes a small but important change to the _init_text_completion_model function in langchain_initializer.py. The change ensures compatibility with text completion APIs by removing the unsupported stream_usage parameter from the kwargs dictionary before initializing the provider class.

TODO:

ensure current implementation for token usage tracking does not break for other providers see #1264

[codecov-commenter]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 69.79%. Comparing base (ef97795) to head (1ec9fc6).

Additional details and impacted files

@@           Coverage Diff            @@
##           develop    #1285   +/-   ##
========================================
  Coverage    69.78%   69.79%           
========================================
  Files          161      161           
  Lines        16057    16058    +1     
========================================
+ Hits         11206    11207    +1     
  Misses        4851     4851           

Flag	Coverage Δ
python	69.79% <100.00%> (+<0.01%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
nemoguardrails/llm/models/langchain_initializer.py	89.51% <100.00%> (+0.08%)	⬆️

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[Copilot]

Pull Request Overview

This PR removes the unsupported stream_usage parameter from the kwargs before initializing text completion providers to prevent API errors.

• Strip out stream_usage so text completion calls don’t receive an invalid argument.
• Ensures compatibility with providers like OpenAI’s AsyncCompletions.

Comments suppressed due to low confidence (1)

nemoguardrails/llm/models/langchain_initializer.py:258

• Add a unit test to verify that passing stream_usage into _init_text_completion_model no longer raises an unexpected keyword argument error, ensuring this change is covered.

    kwargs.pop("stream_usage", None)

[Copilot]

[nitpick] Consider extracting the "stream_usage" key into a shared constant or documenting it in the function docstring to avoid magic strings and make the removal rationale more discoverable.

Suggested change

	kwargs.pop("stream_usage", None)
	kwargs.pop(STREAM_USAGE_KEY, None)

[trebedea]

This is a good fix for now, but we should try to find something more generic for unsupported attributes in kwargs for Langchain LLM providers. Let's merge this and talk in private about a generic fix.