In the quantize_model function, which is called from quantize_static, the first part deals with registering the tensors for quantization and the second part with the actual quantization of each tensor, but the meaning of the function names in the first part is the actual execution of quantization.
For example,
CreateQDQQuantizer(onnx_quantizer, node)
mismatch: this is a factory that selects and returns a per-node handler used in the registration phase, it does not perform quantization. The name suggests creating a “quantizer” that actually quantizes.
quantize_activation_tensor(self, tensor_name: str)
mismatch: this method only registers an activation tensor for later quantization, it does not insert Q/DQ nodes or modify tensor values. The name implies performing quantization of the activation tensor.
and so forth.

https://github.com/microsoft/onnxruntime/blob/main/onnxruntime/python/tools/quantization/qdq_quantizer.py#L551