Octagon not yet supporting BN

[chengchihhong]

In the static analysis engine, with the start of octagon, is written that it will always fetch the weight.

for layerIndex in range(1, len(net.layers)+1):
# in neuralnet.py for storing weights, the index starts with 0
weights = net.layers[layerIndex-1]["weights"]
bias = net.layers[layerIndex-1]["bias"]

This should be adjusted such that it takes different values based on different layers.