Disable document input for AlibabaProvider (Chat Completions API)

[dsfaccini]

• Fixes Input PDF Data into qwen3.6-plus #5724

What

Sending a document (BinaryContent(media_type='application/pdf') or DocumentUrl) to AlibabaProvider + OpenAIChatModel failed with an opaque HTTP 400, because Pydantic AI emits an OpenAI type:file content part that Alibaba's compatible-mode Chat Completions endpoint rejects.

This sets openai_chat_supports_document_input=False on the Alibaba OpenAIModelProfile, so document input now raises a clear UserError client-side — mirroring the existing AzureProvider treatment (#4048).

Verification

The limitation is confirmed by the API's own 400 in #5724 (enumerates the supported content-part types — file is not among them):

ModelHTTPError: status_code: 400, model_name: qwen-long, body: {'message':
"Invalid value: file. Supported values are: 'text','image_url','video_url' and 'video'.",
'type': 'invalid_request_error', 'param': "'messages.[0].content[1].type'", 'code': 'invalid_value'}

Checklist

• Any AI generated code has been reviewed line-by-line by the human PR author, who stands by it.
• No breaking changes in accordance with the version policy.
• PR title is fit for the release changelog.

[dsfaccini]

hey @InfernalAzazel ! could you try this PR's branch's pydantic-ai installation to check whether it solves your issue(s)? I don't have an alibaba cloud setup to test with.