Normally the PP makes the bigger difference, but I find that over hot planetary surfaces, the thruster difference can be huge too. But that's the caveat-- it needs to be a hot environment before the effect seems large. Try testing your builds at Skardee 1, which orbits inside the star's scooping zone. You can overheat any ship there. But some of the cooler running ones, with the coolest possible engineering, can still take some crazy maneuvering to overheat. And by contrast, some ships will always overheat on arrival or departure no matter how much you engineer them. But how many heatsinks you need can vary too.
I haven't tried it with the Dolphin, but it was possible to get the Beluga to manage it with Low Emissions 5 and Clean Tuning 5. By contrast, with Armored 5 on the power plant instead, even with thermal spread it would still run quite hot.
EDIT: These days I usually use Clean Tuning 5 + Drag Drives on most of my exploration builds (or Drive Distributors on the smaller ships, if that gives better performance). The thruster performance with that combination is about half-way between Clean 5 and Dirty 5 without experimental effects, and still runs cooler than stock. If it's one of the hotter ship types, then I might do something other than drag drives, and just take the mild performance loss. The only time I use Dirty is if the ship is particularly cool running already, such as the Krait(s). If I do end up using Dirty, then it gets Thermal Spread.