Use default line colors for better contrast and distinction of auxiliary plots

[axelboc]

Part of silx-kit/jupyterlab-h5web#82

The --secondary-dark color gets a bit lost if I try to give auxiliary plots more distinct colours. Changing to the default colour for the main plot, midnightblue, makes it pop out more, which means I can also use the default aux colours: orangered, forestgreen, crimson, mediumslateblue, sienna.

We've had that green for a long time, so it's going to take some getting used to, but it's not shocking. It's also closer to Matplotlib I think (that's where we got the default colours from, if I recall).

[loichuder]

Me likes it ✨

[loichuder]

Here is the aspect on the dark mode for the record:

I find the colours less distinguishable than on the light mode but maybe that is just my deuteranomaly 🙈

[axelboc]

I've managed to implement dark mode support for the line vis colors. I use data-keep-canvas-colors to stop inverting colors and let users define their own light/dark mode colors.

If the user doesn't specify colors, I pass my own defaults (either one value used for both light and dark, or two separate values) to useCSSCustomProperties, now renamed to useCustomColors. The hook then checks whether dark mode is active and returns the corresponding default color.

Of course, I've then reviewed all the default colors so that they are now consistent and contrasting in both light and dark mode:

[axelboc]

/approve