add a new zebrastack.sln, to replace LeastSquares.sln

https://blog.keras.io/building-autoencoders-in-keras.html

Implementations of the network are at:

• anat0mixer/zebrastack_covidnet.ipynb
• keras_autoencoder/vae.py

Prepping the data at

• anat0mixer/prep_nih - a powershell notebook
• anat0mixer/imagegen.ipynb - tf generator
• dynamic_routing_encoder.ipynb
• anat0mixer.ipynb

Other (non Cxr8) data

• fashion_mnist.ipynb

keras_autoencoder\build_autoencoder.py:from keras.models import Model
keras_autoencoder\keras_autoencoder.py:from keras.models import Model
anat0mixer\dynamic_routing_encoder.ipynb: "from keras.models import Model\n",
anat0mixer\zebrastack_v0_covidnet.ipynb: "from keras.models import Model\n", anat0mixer\zebrastack_v0_covidnet.ipynb: " encoder = keras.models.load_model('data\zebrastack_v0_covidnet_encoder_model')\n", anat0mixer\zebrastack_v0_covidnet.ipynb: " decoder = keras.models.load_model('data\zebrastack_v0_covidnet_decoder_model')" littlecarmash\model_vae_3stage.py:from tensorflow.keras.models import Model littlecarmash\LittleCarVae-cardb.ipynb: "from keras.models import Model\n", littlecarmash\LittleCarVae.ipynb: "from keras.models import Model\n",