Precisely this! There's no need to code the same information twice; it just means you have to think more about what the graph shows.

In the examples, the colour scale improves nothing. The advice is still good in general. A single variable representation can be improved by using a colour scale compared to just gray levels in a lot of cases. And in these cases, using a bidirectional scale for unidirectional data is bad, and vice versa.

I think the "examples" are better understood as displays of the properties of the colour scales, not examples of how they are used in the wild.

That is to say they aren't examples of when you would use a colour scale at all.

He is explaining the situation using the length of bars. He doesn't recommend using both at the same time.