[Feature] Make target optional in Processed data.

[mlahariya]

We should have target as optional input in ProcessedData. Right now target = int. For testing trained models, target is not always available and it should be None.

Eg. we have a trained model M, and we want to evaluate it for a new graphs. We will have to do something like

pd = prepare_processed_data(graph)
target_predicted = M(pd)

Now here - pd itself will have the real target value - which seems very counter intuitive when we are predicting something. Some can claim that the model M does only the follwoing

Def M(pd):
      return pd.target

[Yoric]

I have a PR nearly ready for this.

[Yoric]

Solved by #36 .