Some visualization method names and plot labels are misleading

[zakv]

In particular:

• GaussianProcessVisualizer.plot_hyperparameters_vs_run()
  • This method actually plots fitted hyperparameter values as a function of fit number, not run number, so the name is a bit misleading. The x-axis label and docstring also say run number instead of fit number.
  • The title of the plot accurately says '...vs fit number'.
  • There is one fit per generation, so alternatively you could say it plots fitted hyperparameter values vs generation.
• GaussianProcessLearner.plot_noise_level_vs_run() has the same issues as plot_hyperparameters_vs_run().
• NeuralNetVisualizer.plot_losses()
  • The loss is recorded after every 10 epochs.
  • The docstring, x-axis label, and title say its as a function of "training run", but generally there are more than 10 epochs per training run.

To make things more clear I was thinking of making the following changes:

• Rename the GaussianProcessVisualizer methods mentioned above to end in vs_fit().
  • Update the x-axis labels and docstrings to say fit number.
  • Mention that there is one fit per generation in the docstring.
  • Keep the methods with the old "vs_run" names for now. Make them issue a deprecation warning then call the methods with the new names.
• Change the docstring, x-axis label, and title in NeuralNetVisualizer.plot_losses() to say epochs instead of training run.
  • Change the indices on the x-axis to be 0, 10, 20... instead of 0,1, 2...
  • Mention in the docstring that the loss is recorded every 10 epochs.

@charmasaur do you have any opinions on this? If not I'll just make the changes listed above, probably some time in the next week.

[charmasaur]

All sounds good to me. Sorry about the issues :(

[zakv]

Haha no worries, this one's an easy fix