onnxruntime

32d883ab - Address review comments for WebGPU QuantizeLinear/DequantizeLinear

Address review comments for WebGPU QuantizeLinear/DequantizeLinear QuantizeLinear: Add CacheHint(quantization_type) to prevent shader cache collisions between per-tensor/per-axis/blocked variants. Add T3 type constraint for opset 23+ kernel registration. DequantizeLinear: Add T3 type constraint for opset 23. Add ORT_NOT_IMPLEMENTED guard for output_dtype attribute. Use C++17 nested namespace style. Add const to member variables. Batch uniform variable calls. Fix typo scaler->scale.