The Python Watson SDK attempts its own media type detection when the user provides a file, and it should not

[ry0ohki]

If a user provides a file, but doesn't specify the media type, when adding a document to Discovery using the Python Watson SDK, then Python attempts to auto-detect the media type. However, Discovery itself is capable of determining the media type, and arguably, it has a robust algorithm to do so. As such, the Python SDK should not attempt to do this on its own. Instead, it should default the media type to application/octet-stream which ensures that Discovery's media type detection will be used.

[jsstylos]

Good idea, should be fixed in b0904e723e220d38dda62b173ebb7bb771212ffb