Ironically, regarding RF1 and RF2, I found RF2 to be more resillient. In fact, when I moved to RF1 the mic would drop out once or twice in the first 60 seconds, swapping back to RF2, but keeping the channel number the same, the mic didn't drop out at all.
The channel scan is unreliable in that performing a scan 30 seconds after a previous scan will show that some previously empty channels now have traffic on them; that's the downside of 2.4ghz, and particularly routers that channel scan. In the end, I switched the mic off, did a channel scan, chose one that had no traffic, monitored the red LED's on the RF meter for a while, and then when I was happy it was relatively free I turned the mic back on and set it to that channel.
I'm going to repeat the RF1 and RF2 test after the above setup, see if it makes a difference now. Also, I have a pair of the 180 paddles but I've tested those on stage yet, I'd rather do a long rehearsal in a busy area to see how stable it all is.