I have been thinking about this, bought some parts but have been distracted by other things.
My plan was to abandon the bpf, replace with a broadband splitter like in the qrplabs receiver, use a qrplabs relay board to switch kow pass filters and adopt the class e pa circuit from the kd1jv soda popa which has the similar architecture of the qcx tx path, ie si5351, buffer, 3xbs170 and uses the same inductor coil for each band. So the only switching is lpfs. May need to add a 3mhz hpf before the splitter to keep bcb signals out.
This article on the tayloe detector suggests it may be possible to run with bpfs due to inherent bpf selectivity / high q of the detector.
Compromises on a fully switched bpf, lpf, pa coil approach could include power output, opposite sideband rejection, receive quality with benefit of simplicity, size.
I have read on the forum that the code may not set phase of the iq signals correctly for big freq jumps. Dont know about this.