Erratic Smart Radiator Thermostat Temp Readings

[astephon88]

I've recently picked up a Gen1 HubR, a room thermostat and several radiator thermostats. The control of the room with the room thermostat has been stellar; however, the rooms with only radiator thermostats are struggling to get accurate/sensible readings. I have the Wiser system paired with my Home Assistant instance, so I've been able gather quite a bit of data. Each room also has a Zigbee temp sensor paired with home assistant that I've calibrated with a traceable thermometer, so I'm pretty happy with the readings they give me.  When the radiator valves are open, the temp reported by the iTRV is usually 2 to 2.5 degrees above the actual room temp. I've seen lots saying just make the set point 2 degrees higher, which is ok when valve is frequently opened, but doesn't work when I lower the temp by a few degrees over night. In that case, the iTRV and actual room temperatures  eventually converge, and the artificially boosted set points means the the thermostat will begin heating 2 degrees higher than it needs to. I suppose the overnight set point could be at the actual desired temp while the daytime setting would be elevated by 2 degrees, but this again falls apart e.g. on an exceptionally cold night (where the radiator is warm most of the time, so the readings will be high) or on warmer days (where there isn't much demand for heat, so the reading is actually close to the actual temp). I've also noticed some odd behavior in the temp readings. Sometimes, there will be a prolonged period of steady demand and the iTRV's reading steadily increases. Other times, the valve will open for 2 minutes, and the reading will spike by a degree and stay there for almost an hour, keeping valve from demanding heat again for quite a while. Meanwhile, the actual room temperature continues to fall because of it. The attached plots demonstrate all of these behaviors. During this time period, the boiler is being controlled in relay mode (as a gas boiler) with a 60 degree flow temp. I'm seeing the same behavior with Opentherm, but I wanted to eliminate some of that complexity while debugging.

Has anyone found a solution for this save needing to spend a considerable amount on a room thermostat for every room?