Fix: None case handling in response.output parsing

[JenuelDev]

This pull request makes a small but important change to the response parsing logic to improve robustness when handling API responses.

• Improved error handling in parse_response within src/openai/lib/_parsing/_responses.py by ensuring the code safely iterates over response.output even if it is None, preventing possible runtime errors.

[iqdoctor]

Thanks for the quick fix. This is a duplicate of the earlier PR:

#3315

This PR applies the same parse_response() guard for response.output is None. #3315 already includes that guard and additionally covers:

• regression coverage for the parser path
• Response.output_text with output=None
• streaming ResponseStreamState preservation so an empty/null terminal completed response does not discard already-accumulated streamed output

I compared the diff and did not find extra behavior here that needs to be pulled into #3315.

[JenuelDev]

Going to close this PR then.