Fix: filename not provided in chatcmpl_converter.py

[Nifury]

When a file is provided in base64 encoding, only the file_data part is converted to the chat completion API format.
However, now without the filename field, OpenAI returns
openai.BadRequestError: Error code: 400 - {'error': {'message': "Missing required parameter: 'messages[1].content[1].file.file_id'.", 'type': 'invalid_request_error', 'param': 'messages[1].content[1].file.file_id', 'code': 'missing_required_parameter'}}

[seratch]

Thanks for fixing this; I've confirmed this is a bug and your change resolves it.

[ant31]

The error was already explicit.

openai.BadRequestError: Error code: 400 - {'error': {'message': "Missing required parameter: 'messages[1].content[1].file.file_id'.", 'type': 'invalid_request_error', 'param': 'messages[1].content[1].file.file_id', 'code': 'missing_required_parameter'}}

Openai-agents is not exclusive to GPT.X/openai models, and many models are not requiring a filename to works with file_input.
This change enforce the to add filename, breaking existing and working code with models not requiring it.

if the model used is requiring filename, then error is already there and explicit and the enduser knows what to do.