Add the neural network configuration file for GoogLeNet

[beomyeol]

GoogLeNet, described in Going Deeper with Convolutions, is a 22 layer deep convolutional neural network that won in ILSVRC14 classification and detection. Like AlexNet, GoogLeNet also has been widely used in research related to computer vision and machine learning. Thus, we should add the neural network configuration file for GoogLeNet.

We can use GoogLeNet Caffe reference model as a reference while writing the neural network configuration for GoogLeNet.

[beomyeol]

In order to support GoogLeNet, we should address DAG support (#573) first.