I think that the underlying API of OpenCV changed or else the model itself did, as I ran into a similar problem with scaling the input blob to a DNN recently. I will look into it further.

Hi, I found this code in opencv/dnn/dnn.cpp

if (ddepth == CV_8U)  
    {
        CV_CheckEQ(scalefactor, 1.0, "Scaling is not supported for CV_8U blob depth");
        CV_Assert(mean_ == Scalar() && "Mean subtraction is not supported for CV_8U blob depth");
    }

I think this is the reason for the error.

so, I try to change CV_8U to CV_32F, this code is worked fine.

// fix code
img.ConvertTo(&img, gocv.MatTypeCV32F)

blob := gocv.BlobFromImage(img, ratio, image.Pt(300, 300), mean, swapRGB, false)

Seeing this issue as well. Even converting doesn't work for me.
img.ConvertTo(&imgBlob, gocv.MatTypeCV32F)
blob := gocv.BlobFromImage(imgBlob, 1.0, image.Pt(imgBlob.Rows(), imgBlob.Cols()), gocv.NewScalar(104, 117, 123, 0), false, false)

gives
libc++abi.dylib: terminating with uncaught exception of type cv::Exception: OpenCV(4.3.0) /tmp/opencv-20200531-90131-i8hdtq/opencv-4.3.0/modules/dnn/src/dnn.cpp:334: error: (-215:Assertion failed) mean_ == Scalar() && "Mean subtraction is not supported for CV_8U blob depth" in function 'blobFromImages'