Adjustable vector sizes in CSM plots

[vhirtham]

The current sizes of the coordinate system vectors are fixed to the length of 1. The result is that they are often either not visible (both rendering backends) or distorted (matplotlib if axes are not set to equal). The suggestion is to make the vector size adjustable by the user.
Furthermore, an algorithm that sets the default size depending on the current axis limits should be added.

[CagtayFabry]

from my part we can focus on the k3d implementation and let matplotlib rest at the current state
adding a simple length-scaling factor (with the default being 1 mm) should be enough

[vhirtham]

The thing is, that our online documentation can only render matplotlib images. While jupyter-lab is the recommended working environment, I think we should still update the matplotlib implementation to give halfway decent results. We can't expect everybody to use jupyter-lab and if we want to convince as many persons as possible to use WelDX, the pure Python functionality should offer sufficient support to work with WelDX.

That said, I think that an axis range-related scaling of the vectors shouldn't be that hard.

[CagtayFabry]

yes I'm sure it is simple (and similar) to implement for both mpl and k3d 👍

I honestly haven't used the matplotlib 3d backend in a long time, especially for any "real" dimensions

[vhirtham]

I mainly use it for the tutorials (because I have to) and it is kind of annoying if the plotted LCS are skewed. Since the tutorials are also a part of our "showcase" I want them to look nice without any graphical glitches 😊