Alright, here are some updated results and, YES, I was a dummy. That's what learning is all about, right?
After doing some thinking about it, I realized that when using any of the normal analog inputs, such as the front panel Aux input, and then running the line from the computer and REW into that, even set on "Stereo", the receiver is using bass management to cross from the fronts to the subs. This is obviously not the same as the settings for what the crossover freq for the LFE channel is (which is set to 120hz). So, with my AVR setup to cross to the mains at 80 hz, the sub(s) were being crossed there as well when using anything but a multichannel input that has a dedicated LFE channel. That is why my raw response looked so funny.
I have been trying to do this wrong since the beginning.
NOW I have the computer running into the AVR directly into the analog multichannel subwoofer input (used for Bluray/DVD). This bypasses all processing AND bass management. The results now make much more sense.
Green is the CORRECT raw response close mic of one of the subs.
The others are some different Linkwitz Transform setups from the MiniDSP. I'm thinking that I'll go with the red one and give that a try.
I'm going to try and calculate my actual box volume, which is going to be difficult for me since they are curved cabs. What's confusing and I still need help with is interpreting the graphs to determine what the real raw qtc is. The red trace LT is using the parameters of qtc as 1.3 and the F as 40hz. I really don't think that the box is that undersized to produce such a high qtc, but that's why I need to calculate.
When looking at the graphs, what do you estimate the qtc is (green trace)? Thanks!