Ron, I see your point. And I can certainly understand the temptation to get as much value out of the built-in spectro as possible by treating the HP as an automated patch reader, a vey big DTP if you will.
But in order to measure a profile target with the HP you have to associate it with either a new or current paper, which triggers, among other things ink limiting, maximum density and linearity goals. Now if you choose a new/custom paper, the printer will have to output and measure a calibration target. Obviously this can't come from the desktop Epsons. So then you're measuring a profile target whose L numbers may be way way off what the measurment software expects. What would the ramifications be? How well does the on-board profiling software deal with large variations? Because this is in essence a closed profiling system, I don't know.
My instincts (and this is pure speculation) would say that since the HP system is designed around a calibration step which brings printer behavior into a pre-determined optimum state, that there must be a significant benefit as far as the profiling software goes, to receiving data points within a fairly reasonable tolerance range. Third-party profiling solutions are, on the other hand designed expressly for a wide range of inkjet printer models, in various states of performance. IMHO there are too many potential pitfalls in using the on-board profiling software outside of the uses for which it was designed.
Overly cautious? Perhaps. But we've all got years of experience to remind us how tough it can be just to get products to do what they are advertised to do. So I try to keep my expectations for non-supported behavior to a mimumum .