Make a kente.py Data loader

[robinsonkwame]

Refer to ./src/data/mnist.py, need to make a data loader for kente related project.

• Need to load actual data into X, y, X_test, y_test arrays
• Need to set normal and anomalous class
• What dimension am I working with here? Higher dimension more processing; rescale photos to dimensions.
• What batch size am I working with?
• Do I want to use au-toml techniques? Can I use auto-ml techniques?

[robinsonkwame]

Need to fill this out but a start of a recommend approach, since the code is pretty complex (and somewhat all over the place):

• ~Create a kente.py loader that duplicates minst.py (do the same for the cfg in https://github.com/robinsonkwame/oc-nn/issues/2)~
• ~Update paths, references to Kente folders with minst content.~
• ~Do a trial run to verify that results reproduce (kill when starting to train) <-- Note Google colab will kill your process if you run a lot so I need to be careful~
• Figure out what dimension I need, write a local data loader for that, port
• Figure out structure of where/how normal, anomalous data is read in, converted to dimension, port to class
• Do another trial run, verify it can load (kill when starting to train)
• Do an actual run, fingers crossed
• Assuming results are good, report results, start on paper