Add additional stats to TorchModel training

[alpha-tango]

Currently, TorchModel only prints out:

Epoch 150/150: 100%|██████████| 200/200 [00:10<00:00, 18.58it/s, loss=0.3528, acc=0.819, lr=5.1e-06]
Epoch 150 Summary - Train Loss: 0.3609 | Train Acc: 0.8187 | Val Acc: 0.7937

Whereas HumpBackNorthAtlantic prints out:

200/200 ━━━━━━━━━━━━━━━━━━━━ 33s 166ms/step - binary_accuracy: 0.9642 - fbeta: 0.9362 - fbeta1: 0.9124 - loss: 0.1070 - precision_1: 0.9528 - recall_1: 0.8752 - val_binary_accuracy: 0.2469 - val_fbeta: 0.2907 - val_fbeta1: 0.3960 - val_loss: 16.8190 - val_precision_1: 0.2469 - val_recall_1: 1.0000

It would be nice to have the full set of metrics for TorchModel as well.

[alpha-tango]

Same with BacpipeModel -- only has Accuracy instead of the full stats like HumpBackNorthAtlantic.