Visualise the hidden layer in the model

[Karlo-Emilo]

It should be possible to explore the visualisation through the live data input.

First step: explore if a 2D (or 3D) representation of the latent space in NN gives any meaningful representation.

We should probably use t-SNE: https://distill.pub/2016/misread-tsne/
See this https://colah.github.io/posts/2014-10-Visualizing-MNIST/

Thought: The dimensional reduction could confuse students.

[r59q]

In an effort to keep our changelog history clean, I have scheduled this to version 3.1 in order to close version 2.5 before the domain objects implementation.