All sounds reasonable.
When I’m playing MQA to the DSPs, this may very well be happening: there is no visible indication if it, but it’s a possible architecture. I can see why M would prefer not to send 352 over the MHR wire, but the engineering purpose of splitting it is not clear, if the Speakers can do the upper range, why not all.
But in the case of pre-decoded content, I definitely do not get the second stage indicated on the speakers. If you see that, any suggestions for what is wrong? Firmware upgrade for the speakers?