* support onnx format quantized model

* update code

* add test

* add test

* fix

* fix test

* fix cmake

* update code

* change scale file path to calibration file path

* update code

* update code

* fix build bug

* fix build bugs

* fix

* fix