Construct a TensorFlow model to match the one in `src/shared/pytorch`

The model is defined in `src/shared/pytorch/arch_davenet_orig.py`, `src/shared/pytorch/run_emulator_davenet.py`, and `src/shared/pytorch/network_wst.pkl`. The latter holds the weights and normalisation parameters in pickled format.

Construct a TF model that matches the architecture of the PyTorch model, then copy across the weights.