I use it a fair amount, mostly for your last example type situation, when the 'C' code is more complex than a single line. I find it easier to read to keep the 'loop maintenance' and 'loop actions' code separate.

My most common example is probably a user-input loop: 'B' is while I can read from the input, and 'C' is the prompt/menu. (This assumes an interactive-only program, of course.) 'D' is the actual code to handle the different choices.

Yes, that could be put in the loop in most/all cases, but I like knowing where to look for each type of code, and that they won't interfere with each other.