Measure each driver separately. Be careful to not blow the tweeter out. Measure both woofer and tweeter from a certain distance, for example one meter. Measure the woofer near field ~1 cm. Try to keep the speaker as far as possible from reflecting surfaces, so you can use time-gating. Merge the near and far field response of the woofer together. Measure the impedance of drivers inside the box and import everything into XSim and you will have reliable results.
Ah, I see what you mean.... the electrical response. Yeah, that looks like a bug. As a workaround for now, you could set the drivers's response temporarily to a flat line (same dB all freqs, like go to select and FRD file but then Cancel rather than Open) and mute any other drivers. Won't...
Impedance flattening circuits can be useful with amps of high output resistance but you would need to measure the whole speaker, woofer included. Simulation is just a step to figure out ball park values, this should be confirmed with actual parts and then modified if necessary, to be on the safe...
Low impedance with high frequencies usually increases distortion from the amp, maybe you cold look at trying to increase the impedance a bit on the HP-filter to ease the load on the amp?
Another thing worth exploring IMO is impedance correction of the woofer (upper resonance with BR as you have here). Compare FR at LF of the raw driver with FR with the XO. You usually bump the FR close to resonance with a large coil.
Also compare group delay, that is where the main benefit is IMO. Use a series RLC parallel to the woofer until the impedance is basically flat in the upper resonance. Large inductor is needed, but it can be relatively cheap because DCR is not critical.
Impedance: maybe look at increasing the value of R3 to maybe 10ohm, and then re-adjust the HP values to get the same response, and impedance should be higher. I use R3=15ohm on a DXT tweeter, it's a compromise with what happens at tweeter resonance frequency. I think you are burning a large portion of the HF power in that R3 resistor.
For impedance correction, here is an example: http://www.troelsgravesen.dk/ATS-4.htm
You can also 'hold' the frequency curve with woofer coil shorted, and then put the coil back to 'normal' and compare the FR response.
There is a group delay graph in Xsim that you can open up and compare too with the hold-function if I remember correctly. Minimal group delay is good for tight bass, I also think the GD curve should be as smooth as possible, not like a roller-coaster. I avoid BR because of this, I'm a bit sensitive you might say..
Maybe better if you start a separate thread for stuff related to your speaker, and not to Xsim?