What about using a non-default color to make it clear your doing this so that you use a consistent color through out?

You can use "tab:blue" to get the same color by a fixed name.

yeah so my point is use a color that's not the matplotlib default 'cause that's gonna make it crystal clear that the point of this variable is to keep the color in sync (and the blue isn't an artifact of it being the default)

I like keeping the default colors in the examples. This makes for a homogeneous visual impression and "brand identity". If there is a dedicated color configuration, its correct usage is implied. We don't have to choose another color just for the purpose of proving that this is properly applied.

To be honest, I only made that variable to make it easier to experiment with different colours. I can just as easily remove it since it's only being used in 3 places.

This seems like a dictionary - is there an advantage to this format?

Indeed, you can make this a single dict (problem: list of causes) and pass it to draw_body().

There's no advantage to it. I chose lists over a dictionary because @QuLogic disapproved of that format in the old PR.

I have not followed the whole topic, but my impression was that the no-dict comment refers to passing all data around through all functions as dict.

A dict as the top-level structure defining the diagram is much more concise:

categories = {
    'Method': ['Time consumption', 'Cost', 'Procedures', 'Inefficient process', 'Sampling'],
    'Machine': ['Faulty equipment', 'Compatibility'],
    'Material': ['Poor-quality input', 'Raw materials', 'Supplier',  'Shortage'],
    'Measurement': ['Calibration', 'Performance', 'Wrong measurements'],
    'Environment': ['Bad conditions'],
    'People': ['Lack of training', 'Managers', 'Labor shortage', 'Procedures', 'Sales strategy'],
}