A few more thoughts.....maybe stick a pressure gauge in one rad's blank vent to ensure static head is same as filling pressure of 1.5 bar or whatever, allowing for difference in pressure gauge elevations. If OK increase the filling pressure to 2.5 bar temporarily which might collapse/shift any entrained air.
It does point to a head/circulation problem for whatever reason as one rad at the time only circulating, suggest re installing the UPS3 and beg/borrow a energy monitor, put the pump to full speed 3 and measure the power.
27W = 0 circulation, 36W = 0.3 m3/hr, 5 LPM (one rad maybe circulation). 43W = 0.6m3/hr, 10 LPM (minimum, "normal" expected) and 60W = or > 1.1 m3/hr, 18.33 LPM.