Well, the temperature thing is secondary to the cats, I agree that the cat is probably the main variable. The temp is actually best moderately by something akin to a moderate drive to the emissions shop, followed by the emissions test, then setting for a bit, then another drive... Pretty much just what you did. When you first start the vehicle, the cylinder head temp starts to climb first, (and fast), followed shortly by the coolant. Gradually the temps come up and bring the block along as well. The oil usually takes the longest to change temp. This would likely have all happened on your drive to the shop. The actual test (if it is like the dyno test formerly used here in CA), loads the engine and generates a fair amount of heat. After the test, the vehicle is usually parked for a bit while paperwork and so forth is done. During this time the hottest parts, usually the cylinder head and the coolant in it (which will actually heat up at first, since the water pump is no longer creating any circulation, instead it just soaks up the heat from the combustion chambers) will start to cool, by transferring their heat to the other areas. Another drive after this portion further circulates everything and equalizes the temps. This is good from the perspective of engine running and/or emissions, as everything now should be about where it is designed and engineered to be at its best.

The thing is, the driver usually doesn't know any of this, as modern instrument clusters buffer the data coming from the sensors, and display a "normal" reading over a wide variety of actual conditions. From a coolant temp of about 60* to 120* (Celsius, I've long since forgotten what temps things operate at in Fahrenheit...), the gauge will read a steady, middle of the range reading. So a driver might think the engine is warmed up, but it really takes a few cycles of the thermostat opening and closing before everything is really where it ought to be. And of course, depending on your ambient conditions, this might not happen at all. It isn't uncommon for me to see vehicles with no problems at all, not reach true operating temps without some sort of manipulation f the ambient temp is under about 40* or so (in Fahrenheit, oddly enough, I think of mechanical operating temps in C and weather temps in F...).

But this is all secondary, my money is on the cat(s) being hotter and/or "lit off" for the second test. To answer you main question though, I've found that the emissions tests do a fairly decent job of producing repeatable results. If things are changing drastically in the test results, things are usually changing on the vehicle.

MV