Fix Torch2Trt in Savant 0.6 extra

[denisvmedyantsev]

model = resnet.resnet18().eval().cuda()
dummy_input = torch.full((1, 3, 224, 224), 0.5, dtype=torch.float32).cuda()
model_trt = torch2trt(model, [dummy_input])

> AttributeError: 'tensorrt.tensorrt.INetworkDefinition' object has no attribute 'add_convolution'. Did you mean: 'add_convolution_nd'?

We have a sample with torch2trt - nanosam, see #1110

Consider replacing torch2trt with Torch-TensorRT