Ok, as you seem to be more educated on the matter than myself, please correct my equation. Nothing comes for free.
220 and 250 have the same mass, so need the same amount of fuel to do the same performance.
The 250 will use more fuel than the 220 if you are using 100% performance.
Of course manufacturers go to great lengths to reduce the g/km, obviously not just chance.
Some new volvos sneak into the lowest band just by loosing the weight of the spare wheel and associated kit. If you option the car with spare or with certain larger wheels with greater unsprung mass they end up in the higher tax bracket.
The thing about theory is that it doesn't always transfer (intuitively) into the real world. Take cruise control for example; many may think that liberal use of cruise during a journey will help improve economy, but it actually worsens economy due to the computer constantly varying the fuel supply to maintain speed. Some 'theorists' will argue against this phenomenon, but it is true in practice.
Converting diesel into energy is similarly counter intuitive at times and complex. I guess most will accept that if they have a 20 year old gas boiler, it will not be as efficient as a properly set up new one. Just adding more fuel doesn't automatically mean more heat and energy.
Of course modern computer controlled diesel engines do a wonderful job of getting it right and producing 'near' maximum combustion efficiencies (as available from the hardware) 'most' of the time.
But to do this reliably across the rev range at all temperatures, would require (for example) fuel that was consistent within a much finer tolerance that from your average ASDA. You could go on of course, same wheel, tyres, tyre pressures, weight of passengers, humidity, etc etc.
Then we go onto the need for the manufacturers to get highest possible combined mpg, and lowest possible CO2. They can do this by preparing a map that is tailored to ensure the leanest possible mixture at the revs/points in time that are critical to the tests. This tends to result is the less smooth performance through the rev range than a stock car can exhibit, and one that is often 'cured' by a remap.
The gap between official mpg/CO2 figures and real world are getting wider and wider. It's now common for a 65 mpg official figure to see only 45 mpg in the real world. This is a further sign of how manufacturers are tuning to pass a test, not produce the 'best map'.
As highlighted before, many engines are effectively 'detuned' (solely by means of the map) by the manufacturers to meet marketing/legislative/business criteria. Surely this is absolute proof that remapping is a very real and useful tool.