Support code_interpreter and vision for file uploads

[bruffridge]

Deep-chat 2.0 added support for v2 Assistants API. It also changed how uploaded files are handled. Now there is a files_tool_type parameter that determines which tool operates on file uploads. So instead of letting OpenAI decide all uploaded files are handled by the tool set by the parameter. This means ALL uploaded files can only be handled by one tool. We want it to be based on the file and what the user asks. For example, uploaded images should be passed to the gpt-4o for vision. Other files such as excel and csv the user might want visualized via code_interpreter. PDFs may want summarized via file_search.

OvidijusParsiunas/deep-chat#190 (comment)

[bruffridge]

Can workaround by defining a function for files_tool_type that returns the tool based on the file type. Waiting for OvidijusParsiunas/deep-chat#281

[OvidijusParsiunas]

Hi @bruffridge.
The issue mentioned above should already be taken care of. Hence you should be able to use the files_tool_type property as a function without any problems.

[bruffridge]

Fixed in the azure branch.