Add endpoint on the client to retrieve the stats of the last training

[eyp]

Don't know what should be the best approach, but it'd be nice to retrieve, for example the evolution of the accuracy metric on every epoch of the training:

Accuracy of model trained at epoch 1 : 0.9118
Accuracy of model trained at epoch 2 : 0.9118
Accuracy of model trained at epoch 3 : 0.9118
Accuracy of model trained at epoch 4 : 0.9118
Accuracy of model trained at epoch 5 : 0.8824
Accuracy of model trained at epoch 6 : 0.8824
Accuracy of model trained at epoch 7 : 0.9118
Accuracy of model trained at epoch 8 : 0.9118
Accuracy of model trained at epoch 9 : 0.9118
Accuracy of model trained at epoch 10 : 0.9118
Accuracy of model trained at epoch 11 : 0.9118
Accuracy of model trained at epoch 12 : 0.9118
Accuracy of model trained at epoch 13 : 0.9118
Accuracy of model trained at epoch 14 : 0.9118
Accuracy of model trained at epoch 15 : 0.9118
Accuracy of model trained at epoch 16 : 0.9412
Accuracy of model trained at epoch 17 : 0.9412
Accuracy of model trained at epoch 18 : 0.9412
Accuracy of model trained at epoch 19 : 0.9412
Accuracy of model trained at epoch 20 : 0.9412

instead of retrieveing so much text, just something like:

[{epoch: 1, accuracy: 0.9118}, {epoch: 2, accuracy: 0.9118}, etc}]

Probably good idea is to add also the time spent on this training, or the time spent in each epoch.

Of course, it must be added some kind of persistence (memory or database) on the client to be able to store the data of the last training.