Hello, I read -most- of the posts here and in TRV Zigbee Thermostatic Radiator Valve - implementation of scene functions for correct control and would like to give my 2 cents.
First of all, I was “stopped” by this affirmation:
don’t get me wrong, I am NOT saying it is wrong or anything, I come from a country that doesn’t need heating systems in every house, lived 9 months in Southampton and not I am dealing with these things in the house where I live in Germany…
My house has a Junkers Boiler with its control (a little more modern than your Honeywell in the picture) and the “possibility” (not installed) of having a water tank for the warm-water supply, so “similar” to what you describe.
The only Radiator that has a “normal” valve is at the entrance, therefore I wonder if it has to do with what you mention (“emergency heat dump”).
But, the installation manual of the Junkers boiler and its control say that the control should be installed in a room in which the radiators DO NOT have thermostatic valves, but normal ones like the one in that radiator at the entrance.
There is a “good” reason for this, it has to do with the “control loop” for each thing. The Thermostatic Valve in itself is a control loop, the easiest one: “below a temperature, open valve, above a temperature, close valve”, with the hysteresis (“reaction delay”) provided b the system (the time for the extra hot water to heat the metal and to radiate in the air to the 5cms away were the thermostat actually is)
In this loop, there is one input (“set temperature”) and one output (open/close valve). We tend to think that we are controlling the temperature of the room, we are not, just on that point 5 cms away of the radiator. The moment we put the temperature further away, we change the hysteresis part of the loop.
What I found in my house: The Boiler control is in the living room (inside wall) and there is a very large radiator in the opposite wall under the window with a thermostatic valve with an extension-temperature probe which could be hung further. So right now it would be measuring the temperature before the Xmas Tree if I had not replaced it with the TRV. There is a second small radiator but its Valve is not working, so it is always off.
My “distant future” plan is similar to what others mention here, but since the Junkers control is analog (0-20V) it would be a little harder. Actually I measured the output of the controller and seams to have only this possibilities: about 5V: boiler off, above 15V: boiler full power, and two intermediate around 9 and 12V (not sure if it is the same) for middle power. Hacking this part will take me a while.
But supposing I already have the control signal, the decision for that control signal comes from a combination of temperature sensors. At this point, I wouldn’t need to control the valves of the thermostats, they should be controlled by their own individual loops, except maybe in the radiator in the same room of the Main Control of the Boiler (following the installation manual). The decision around these 3 or 4 possible settings for the boiler (including off) would come from a combination of all the thermometers in the house. Here I would NOT want to use the temperature of the thermostats, 5cms away from the radiator (that is my biggest conflict with these heater system! It is too close to be measuring the “stable state” of the room, but it works almost the same for “any room” where it could be installed). I would have a thermometer in the “most relevant” position of each room. I just put one under the junkers controller, since it doesn’t display the temperature and I suspect it is not calibrated, I have the settings in 30° for getting 20° 
So I would somehow make a “weighted average” of the temperatures in the different rooms (the weights would be a little empirical, including room size, if the room has windows, importance of the room being stable, etc), then decide how far away of my “expected value” is that, in order to choose the power to sent to the boiler. For those boilers with just an on/of dry contact its easier at this point.
so, as I just saw from updating eWelink to 5.2 (after reading the other thread) it includes as trigger if the TRV is heating or inactive, that is enough information from the TRV (as discussed in the other thread, if all the TRVs are inactive, the boiler can be turned off, if at least one is “heating”, the boiler should be turned on, and in my case for “several” TRVs heating, put full power!). I believe the “local” control should be left to the TRV control loop.
With this 5.2 update it should be possible to do what everyone wanted, to use the temperature of the thermostat for changing other things, including the temperature setting of the same thermostat??? It wouldn’t make sense but it is the only way to force open/close the valve, and I don’t expect to ever have a direct open/close valve control since that would be messing with the control loop of the thermostat itself…
Anyway, I just wrote what I have been “thinking out loud”, hope it gives you ideas.