And please do keep enjoying your SubMersive HP.
The point I believe you are missing in the above discussion is that we are only tweaking the settings made by Audyssey which are effectively assumptions without any verification on Audyssey's part. You reference trade offs over the listening area. If we were adding post EQ to the subwoofer channel, then you might have substance for a complaint, but we are leaving all of the processing Audyssey has done to the response of each speaker channel. As I recall, we are instructed to carefully pick the first measurement location, as this *single location* is used for delay settings. As such, the delay is not optimized for all locations in the way the speaker correction is.
If we were setting up our home theater in a large back yard this delay calculation would be very simple and straight forward. Within a home theater and with real loudspeakers picking a precise delay time is anything but straight forward if you are looking at anything below 10kHz. The "automagic systems," as Dennis Erskine calls them, must make a best assumption on what will most often be relevant to what we hear to do their calculation. Getting a specific number with a few decimal points at the end doesn't mean the number is more correct. To steal a quote my Dad often uses, it's generally better to be reasonably correct than precisely wrong.
This also heads into the reason for delays to exist in a surround system at all. The early processors only offered delay adjustments (in milliseconds) for surround and center speakers. Later they realized it is a trivial calculation to allow users to instead set up delays by entering the measured distances to each speaker. Since the functions were already there it only made sense to include the subwoofer in the delay functions. The purpose of the delay adjustments for main speakers is to aide with multi-channel imaging and surround field envelopment. With the subwoofer being a single channel, it has no imaging effects for which this is needed. The delay setting for the subwoofer channel is there to keep the subwoofer in relative sync with the arrival of sound from the main speakers.
If the subwoofer's sound reaches the listener much sooner than the overlapping sounds from the main speakers (ie crossover range), it will call much more attention to itself and can cause localization issues. Arrival differences also affect how closely the crossovers actually function to the theoretical ideal. Unfortunately the theoretical ideal can only work ideally for a single location. When the arrival times for a given frequency range change, the summation between the two signals change and can range from a maximum of +6dB reference to either, or a deep cancellation of 20+dB. For reference, an 80Hz wavelength is just over 14' long. If we had a perfect 6dB summation at 80Hz and then added a 7' equivalent delay, we would then see a very deep notch at 80Hz. In the real world we are dealing with the interaction of many frequencies, where 1-6' can have quite significant effects on the resulting interaction.
I fully expect this aspect of auto-optimization will see continued revisions and improvements to the measurement, calculation and confirmation process. For a very simplified analogy, just because your favorite GPS Navigation system doesn't know that a bridge is removed for construction, I wouldn't expect you to drive off the bridge it directs you to take. In the case of audio setup, without some basic measurement systems we can't see out the front windshield that the bridge is out and the assumption of the computer is incorrect.