I have noticed that on a new (pre-release) version of
the software I'd been test-monkeying, the Engine
Delivered Torque param is now looking properly
scaled and responsive. I have used this to observe
flywheel torque output for my new cutout open &
closed, etc. (about 10 lb-ft top end gain). The data
can be a bit choppy on fast runups but useful anyway.
The chop is less than the gains, at any rate, and some
eyeball averaging suffices.

Here is my suggestion... if you log this value, along with
spark, KR, MAP, dynamic cylinder air, RPM, TPS, speed
and O2s, you have basically a rolling dyno. I use the
Interstate, slow down to 70, punch it for a downshift
and let it run back up past the 2-3 shift. This gives a
fairly short (maybe 20-30 ticks) "dyno pull". The other
(non-TQ) params are the main indices for various
tables and key to figuring out where you're operating.

Now, if you set out on a short road trip with a pocket
full of .bins you set up ahead of time, ones with series
of planned-out spark map advances, PE-vs-RPM
differences, and logged each of the different tunes
(loading the next one, every exit) in under an hour
you could have a really good set of data for how much
XX timing advance helped / hurt, how much PE, etc.
as well as more detailed insight (maybe KR comes in
here, and a more surgical PE tweak or spark pullback
above YYY kPa makes more sense given current O2s,
etc.).

As far as response to param tweaks, I saw 330lb-ft the
other day as a peak w/ stock timing table and pushing
the mid-MAP +4, top end +2 (which gives me 6000RPM
advance of only 27.5 degrees, still) I got 360lb-ft and
still no KR. So this looks like a good, strong indicator
and a very usable tuning feedback param. You can
clip out pulls (TQ and RPM) and scatter-plot them to
see a dyno-looking torque curve in Excel. I haven't
worked up the FWHP(FWTQ,RPM) calculation but it's a
trivial one. If you go by torque and just maximize it all
across the range, HP has to follow directly; display is
just a nicety / amusement.