Todos for plotting methods after fleur workshop

[janssenhenning]

These things I plan to work on for the plotting methods but probably won't get to till after the fleur workshop:

• matplotlib has a similar capability to bokeh to provide a dataframe and the keys you want to plot via the data argument (https://stackoverflow.com/questions/39919170/matplotlib-scatter-plot-how-to-use-the-data-argument). We can use that to unify the interface for passing data and provide a unified way to still provide direct lists to plot (look at bokeh_line for example)
• plot_fleur_dos should be able to take kwargs with parameters specified in dicts with the keys corresponding to the DOS component. This way we could say color={'MT:1': 'red'} without needing to know the exact order of the plots
• Implement unified plotting functions for common plots for all backends
• Expand tests in plot_methods and bokeh_plots
• I'm still not happy with the default color palette for DOS plots. Maybe we can design something custom there
• Bokeh bandstructure plots have no working tooltips at the moment. One idea I had for this:
  • Hover tool to highlight complete bands and show avg/min/max eigenvalues and in weighted case show avg/min/max weight
  • Selection tool for bands
• Look how much can be simplified in periodic_table_plot by introducing the BokehPlotter

[janssenhenning]

• Bokeh plots do not natively support Latex Labels. But there are solutions for this in the bokeh documentation it is just a matter of making them obvious to use here

[janssenhenning]

• passing linestyle to plot_fleur_bands has no effect (Problem is that the data is not ordered that a line plot would connect the same band at the moment)

[janssenhenning]

• decide whether constrained_layout should be True by default in matplotlib
• Reference images for matplotlib tests should be easier to update (rethink folder structure)