Add tensorflow utility functions to efficiently feed DL tools

[EiffL]

Images can be generated on the fly and cached using the tf.dataset API, this will be useful for feeding Deep Learning or other TensorFlow-based algorithms.

Here is an example of how to do that with GalSim: https://gist.github.com/EiffL/f7d06252b90581d3b00d01ea466257e6

We would like to be able to provide such tools for the user interface

[ismael-mendoza]

This sounds like a good idea, what do you think @sowmyakth ?

I was thinking that good places to start would be generate in measure.py and in draw_blends.py.

[ismael-mendoza]

Thinking about this some more, I realized it might be useful to add this feature but let it be optional as to not enforce tensorflow or pytorch as a requirement to use BTK although I'm open to suggestions. So we would have an option for the user to work exclusively in numpy and another option to provide a pytorch dataset for instance.

Having an option to provide e.g. a pytorch dataset would lower the barrier users that are training/testing ML algorithms to use BTK.

[ismael-mendoza]

Feel like the generators can be very easily be turned into a pytorch dataset by user so not a priority at the moment.