We can see that Annex D and Annex D3(a) of the 2011 provide very different views of feeder sizing. Why is the Annex D3(a) example so much more detailed and complex? For one thing, in Annex D3(a), you size two different types of feeder: ungrounded and grounded. That is, the current-carrying ones and the neutral.
The complexity begins with sizing the current-carrying conductors. Under the heading "Ungrounded Feeder Conductors," the first thing you read is that the conductors must independently meet requirements for terminations and for conditions of use throughout the raceway run. This isn't an arbitrary limitation imposed to make the example interesting. It's derived from applying 110.14(C).
In any given run, you must size the conductor for the lowest temperature rating of any connected termination, conductor, or device. But what if a run has no connectors? That’s where the “independently” part comes in.
If the circuit must transition a particularly hot area, such as a steam tunnel, you can save considerable expense by addressing conductors and terminations separately. That’s what’s being done in the Annex D3(a) example. It’s not necessary to size the conductors of the entire circuit for that portion that goes through the steam tunnel. You can keep terminations and devices in the cooler temperature area, using a higher-rated conductor in the higher temperature area.