The impact is not on the overall system efficiency, it's in the maximum sustainable power the electronics can handle before hitting their thermal limit.
Take the ST part, the junction to case temperature coefficient is 0.45 C/W, assume 0.55 C/W for mounting and cold plate, total of 1 C/W. The max junction temperature is 175C, if the coolant loop is running at 35C, then there is a 140C delta to work with, that works out to 140W of dissipation.
For the nominal part (at 25C), the sustainable current level is i^2*R=140, with 18 mOhm resistance, that's 88.2Amps. For the worst case, the resistance is 28 mOhm and the current limit is 70.8 Amps.
Now look at what happens as the die heats up: the 175C nominal resistance is 25 mOhm, a 39% increase from 25C and a limit of 74.8A
If that trend holds for the worst case part (Figure 12 indicates a 50% change for 25 to 175), the hot resistance is 39mOhm with a current limit of 60 Amps.
Thus a max temp nominal part has better current/ power handling ability as an ambient worst case part. At the hot end, it provides 18% more current. However, it runs cooler, so for the same power profile it will be at a lower temperature and thus more efficient/ more power available.
So a worse performing part gets hotter faster and even less capable faster. Not a big deal for street use, but for a P on the track, it can definitely affect things.