After quantification, add pipeline operation

## ❓Question

After I trained the yolov8 model, I used coremltools==7.0 to convert it into an mlpackage model. Then, I quantified the model using the pruning in the official documentation. After quantification, I added nms operations to the model using the pipeline mechanism. I imported the model with nms added into xcode and input the predicted images. xcode crashed directly. By importing an nms model without adding quantization operations, xcode could predict the results normally

No source file or test is named. Start by reproducing the coremltools 7.0 conversion, documentation-based quantization, and NMS pipeline flow, then import the resulting model into Xcode and inspect the crash. Done means identifying the quantized-pipeline incompatibility and verifying a working model or documenting the limitation.

Written by the indexing model from the issue text.