Embed radiomics tools from qtim_tools into DeepNeuro as either a "model", or and "output", or both.

[AnnaBeers]

This will be good for those who want to use radiomics as a bench mark for deep learning efforts. A "model" object could extract radiomics features, and then run them through a random forest (or another preferred model, like SVM). An "output" object could just save out the features to a csv file. Given that a radiomics + RF model does not iterate over batches in the same way that a deep learning model does, it might be better to squeeze it all into an "output" object.

This would require a bit of auxillary code, like code to extract regions around a provided region of interest, but this shouldn't be too hard, and would be useful anyway!