As a native english speaker I can think of many ways in which the word 'dummy' could be used. Here are some.
A dummy run. Can be used in many circumstances where an operation is performed with some missing ingredient. For example, in Rugby where a player seemingly heading to score a try adopts body language suggesting he is carrying the ball in his arms when actually another player has it.
A dummy. A pacifier, something you offer a fractious baby to keep it quiet.
Crash dummy. A replica human used when safety-testing cars.
And so on.
I see no objection at all to using the term to describe the testing of a control in a scientific test. Objections are merely a matter of rather desperate use of semantics, not of science. As for assigning such a test a COP =1, it hardly matters, since by doing so you assume the whole system is conforming to the normal thermodynamic rules, and neither creating nor destroying any energy, while ignoring the fact that some of it may be transformed into forms you cannot detect.
So long as the same instrumentation and methods are used for 'live' tests, and we assume that the difference between dummy and live runs is sufficiently large to remove all doubt, it is not absolutely wrong to do this when performing a qualitative experiment. Quantitative experiments with proper calorimetry would be a different matter of course, but we have no way of knowing into which category the current experiments should be placed.
For example, if someone were to design a miracle fuel-saving carburettor for a car, I could see no objection to defining the COP of the car with a conventional carburettor as unity (COP=1) and thus describing any improvement with the miracle carb as being a multiple of that.
This of course, assumes that any of this stuff works at all, a matter on which I am for the moment reserving judgement and some element of proportion.