[Feature suggestion] More than one output nodes, results in multiple models

[piyush-palta]

Currently user can make more than one output nodes, without any consequence. Since user can be total beginner and might not realize what he is doing, will be presented with erroneous codes.
One way to deal with such scenarios could be as suggested in #114 .
Another way could be instead of presenting it as error, we can incorporate this as a feature i.e. When user try to create the second Output node, we can ask him if he wishes to make another model, with same network, just different output node. This way we'll have two different models.
This can be really useful in use cases where same network has to be used for multiple tasks. For eg: One output node can classify image into Non-Living and Living thing, while other output node recognizes image and classify the object itself.
Although in the backend, two different models will be created, this feature would be really useful to get the whole task done efficiently.