Anyone using Cisco FRA in production? Started testing and am noticing a feedback loop with RRM and FRA

So we have started deploying 3802Is on 8.5.120.0 and recently started testing FRA. Set all the radios to static role assignment, turned on FRA, then started turning the radios with a 100% COF to auto role assignment. We started with ~100 APs in a moderately dense deployment. Minimal LoS AP placement, normally a wall or two between placements, but still about ~25-40ft apart.

This initially seemed to go well, first COF run we had 8 APs show 100%. All the 2.4 Ghz radios were running a TX power of ~7-8. Set the 100% COF APs to auto and they converted to dual 5Ghz. Next batch about 5 APs, same story. Rinse and repeat for a couple of hours.

I've let it bake for about 12 hours and come back to something disturbing. We're up to 23 APs that have been converted to dual 5Ghz, and another 10 APs saying they have 100% COF. It seems like RRM and FRA are starting to form a feedback loop. FRA will disable an AP, RMM will boost power of the surrounding APs. FRA will have new redundant APs, convert, and RMM will boost the power.

So far I've noticed at least one area where a coverage hole seems to have formed. The center AP was converted, then later the 3 surrounding APs were converted. For the most part the 2.4 radios are still TX powered between 7-8, but there is a group starting to raise into the 4-5 range. Right now about 80% of the clients are still reporting RSSIs of => -65dBm and 70% are => -60. Signal quality is also looking really good with 90% at => -25dB. Our worst 5Ghz client is also reporting -77dBm, and the worst 2.4Ghz is reporting -66dBm. So the numbers say there isn't actually a problem and I should keep letting FRA go to town.

I've read the documentation, and I know that if there's an actual coverage hole, FRA will dump a radio back to 2.4 immediately. I know I have set an upper limit on what I want my 2.4 radios to TX at. I've looked at the debug and know that I'm not getting any significant bleed between floors. I'm just not sure if I want to ride the train all the way to end of the track where RRM has raised all the 2.4 radios to a TX power of 4, and FRA determined half of my APs were now at 100% COF.

TL;DR: FRA and RRM seem to be feeding into each other, FRA disabling 2.4 radios, RRM increasing TX power. FRA disabling more 2.4 radios. Repeat. Do I really want to end up with a bunch of 2.4 cells with a TX power of 4, and half my APs turned to dual 5Ghz?