Actually iperf3 gives out both. I just really dislike this whole Mbits thing. MByte seems much easier to imagine.
I was running a pair of OpenWRT flashed NanoBridges M2 on the exact same spots for close to a decade and they were constantly getting around 3MBytes/sec (that's 25Mbit/sec for you pedants). It's not the best PtP in the world far from it. It's in an area with tens of other devices (several of which are 40MHz forced) and I don't even have LoS. 3MB/s seemed like an absolute miracle but now I'm pushing twice as much. 
I was just curious to know why is that given the settings are contradicting my measurements.
Well I wouldn't call 2 years of releases "barely" supported. I might try 22.03 once it's released. I'm running it on some other non critical devices already but this PtP is far too important to me to try a release candidate on.
Anyway let's close this mess of a thread on it being a mystery caused by a higher power (a bug in ath9k most likely).