advanced_padding.py import errors fixed by copying keras 2.2.0 (used in TF1 MLAir) modules to mlair/keras_legacy. Result not tested yet.

make_keras_pickable hotfix added from: https://github.com/tensorflow/tensorflow/issues/34697#issuecomment-627193883. It is placed into helpers.py and runs in __init__() of AbstractModelClass to always be called before any model configuration.

deactivated not working advanced_paddings.py import, changed some other keras imports to tensorflow.keras as keras, *.model.fit_generator() changed to *.model.fit() which is said to have the same functionality - Commit suffers from PicklingError with RLocks when trying to save Callback in Epoch 0002

tf_upgrade_tf2 run on whole directory. Only training.py and abstract_model_class.py directly affected by the script. Other problems were mostly import keras related and solved with import tensorflow.keras as keras. Not possible for advanced_paddings.py. For details see: #331

updated all run scripts with new Workflow class, HPC_setup script is calling now the new run_HPC.py script

split experiment path and name setup, replace parser inside ExpSetup by experiment_date parameter and add therefore args handling to all run scripts

introduced module "modules" to collect all working steps (like experiment setup, pre-processing) in separate folder

some paths are still not updated and will cause errors (WIP)

restructured experiment setup, therefore deactivated some functionality in run.py and modules.py (tests fill probably fail now!)

split run.py into experiment_setup.py and modules.py (includes preprocessing, training, postprocessing)