`OpenAIResponsesChatGenerator` drops reasoning `encrypted_content` in streaming mode.

[datbth]

Describe the bug
OpenAIResponsesChatGenerator drops reasoning encrypted_content in streaming mode.

Error message
N/A

Expected behavior
OpenAIResponsesChatGenerator#_convert_streaming_chunks_to_chat_message preserves encrypted_content in ReasoningContent.extra, so that it is submitted in future chat turns.

Additional context

• https://github.com/deepset-ai/haystack/blob/main/haystack/components/generators/chat/openai_responses.py#L815
• https://developers.openai.com/api/docs/guides/reasoning#keeping-reasoning-items-in-context

If the model calls multiple functions consecutively, you should pass back all reasoning items, function call items, and function call output items, since the last user message. This allows the model to continue its reasoning process to produce better results in the most token-efficient manner.

• https://developers.openai.com/api/docs/guides/reasoning#encrypted-reasoning-items
• https://github.com/openai/openai-cookbook/blob/main/examples/responses_api/reasoning_items.ipynb

To Reproduce
Run OpenAIResponsesChatGenerator with generation_kwargs of

{
  "model": "gpt-5.5",
  "store": False,
  "include": ["reasoning.encrypted_content"],
}

and streaming_callback

FAQ Check

• Have you had a look at our new FAQ page?

System:

• OS: Linux
• Haystack version: master

[obielin]

@anakin87 I took a look at this and believe the issue is in the streaming conversion path where encrypted_content is being dropped. Happy to work on a fix if it's available.