Modify convergence plotting

[zStupan]

Summary

• Modified convergence plotting so that a user can control whether generations or evaluations will be on the x axis of the graph.
  In case of generations/iterations the best global fitness at each iteration is plotted. In case of evaluations, we have the number of evaluations at which improvements were made on the x axis and the fitness values at those points on the y axis.
• Renamed task.plot() to task.plot_convergence().
• Renamed task.return_conv() to task.convergence_data().
• Both task.convergence_data() and task.plot_convergence() have an argument x_axis which can be either 'evals' or 'iters' and it controls what goes on the x_axis. The default is 'iters'. task.plot_convergence() has an aditional parameter title, to set the title of the matplotlib graph.
• task.convergence_data() now returns a tuple of numpy arrays instead of python lists.
• Also fixed a flake8 dependency issue which caused it to crash.

[firefly-cpp]

Thanks @zStupan! Can we merge this commit?

[zStupan]

@firefly-cpp Yes

[firefly-cpp]

I ask @GregaVrbancic to release a fresh version of this package -- 2.0.1

[GregaVrbancic]

I ask @GregaVrbancic to release a fresh version of this package -- 2.0.1

Done