This relates to posts #1-4 above.
The EMC failure is bad for us. Our product has no filter at all, not even a decoupling capacitor, ...do you think that the EMC test house saw this, and then secretly but deliberately activated a noisy_mains_supply_injector to try and upset our equipment, and interfere with it, such that it started malfuncationing and then failed EMC?
Apparently they (the EMC test houses) secretly, without telling you, put an injection circuit at the mains input to your equipment. That typically consists of an inductor in series with each line that can carry the equipment current and has a high inductive impedance at the frequencies of interest. Then they use capacitors to inject the noise into the lines between the inductors and the equipment under test.
I am sure the EMC test house secretly do this sort of thing, especially when they get a product which has no filter...do you agree?