fix: pass config params to structured_output in OpenAIResponsesModel

[Di-Is]

Description

OpenAIResponsesModel.structured_output() silently drops all config params (max_output_tokens, reasoning, instructions, etc.) because it only extracts ["input"] from the _format_request() result.

This fix builds parse_kwargs from the full _format_request() output, excluding model/input (already handled explicitly) and stream (hardcoded as True by _format_request(), incompatible with parse()). Other params pass through directly, letting the API/SDK validate incompatible combinations.

Resolves: #1908

Related Issues

#1908

Documentation PR

N/A

Type of Change

Bug fix

Testing

• test_structured_output_passes_config_params — verifies max_output_tokens and reasoning are forwarded, stream is excluded
• test_structured_output_passes_instructions — verifies system_prompt is forwarded as instructions
• Verified with live API calls against gpt-5.4 that responses.parse() correctly accepts max_output_tokens, reasoning, and instructions

---

• I ran hatch run prepare

Checklist

• I have read the CONTRIBUTING document
• I have added any necessary tests that prove my fix is effective or my feature works
• I have updated the documentation accordingly
• I have added an appropriate example to the documentation to outline the feature, or no new docs are needed
• My changes generate no new warnings
• Any dependent changes have been merged and published

---

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.