Support propagates timeout parameter from request options for OpenAi models

[guanxuc]

Description

Support propagates timeout parameter from request options for OpenAi models.

Related issue

Relate to #6604

[seeun0210]

Thanks for taking this on @guanxuc! 🙌

This is the OpenAI counterpart to the same root cause behind #6604 — in #6605 (the Anthropic fix) I flagged that OpenAiChatModel.internalCall()/internalStream() had the identical issue (calling .create(request)/.createStreaming(request) with no RequestOptions), so it's great to see it covered here.

For reference, #6605 fixes the Anthropic side by passing the resolved timeout through the SDK's two-arg create(params, RequestOptions) overload — might be worth keeping the two approaches consistent so reviewers can look at them together. Nice work! 👍