This MR is the implementation of #87.
As of the creation of this MR, it is based on the branch QuantizedONNX, branch created to improve the import and export of ONNX models. See !154 MR of QuantizedONNX