Sam
What you describe is something that I would describe as hand-drawing the final stage (though I am not sure whether Axbridgecaver would).
These things are never clear-cut, however, and Peak is a good example. The thing is that in long and complex systems, new extensions, especailly ones that add new closed loops through previously known passages (or across the surface) do not just
add to the survey but also
distort it. Thus the previous drawing can be made redundant. For this reason, many long and "active" caves (in the sense of ongoing exploration) are frequently presented just by what can be generated by the computer.
What is needed, in my opinion, and has not been addressed by the software pilots that we have is an output file, like a survex plot file, which has additional layers on which the drawing and labelling can be put, as you say John Beck does, but which remain linked back to the data files, so that a modified plot can be dropped onto the "current" drawing (probably with a colour change to indicate changes) so that only minimal changes to drawing and labelling need to be undertaken to keep the main drawing up to date.
Of course different layers could be used for an "overall view" and "close ups" and other variations are possible, and I suspect that this approach might be more fruitful in many cases than current attempst to get the computer to do all the drawing.
Of course I just have the ideas, it's up to the geeks to make 'em work.