Add additional models

[Flunzmas]

This project becomes more and more useful with the amount of models trainable or available (pre-trained):

• CopyLastFrame baseline
• LSTM -> revise implementation
• ConvLSTM -> revise implementation to match original paper
• U-Net with temporal convolutions
• PhyDNet -> revise implementation to match original paper
• ST-LSTM/PredRNN(++) -> revise implementation to match PredRNNV2 paper (https://github.com/thuml/predrnn-pytorch)
• Noether Networks (https://github.com/dylandoblar/noether-networks)
• SVG (https://github.com/edenton/svg)
• MIM (https://github.com/Yunbo426/MIM, PyTorch: https://github.com/bugsuse/MIM_PyTorch)
• E3D-LSTM (https://github.com/google/e3d_lstm, PyTorch: https://github.com/metrofun/E3D-LSTM)
• SAVP (https://github.com/alexlee-gk/video_prediction)
• LFDTN (https://github.com/AIS-Bonn/Local_Freq_Transformer_Net)
• ...