It may be a little off topic, but since you mention it: the higher speeds on a ride don't affect the average as much as the lower ones. Just model it out, start with the record of a real ride if you want and "what if" you increased the higher speeds by half? Then "what if" you increased the lower speeds by half. You'll see that bringing up the slack or resting speeds influences the average, but slamming it downhill does almost nothing for it.

Of course, hammering downhill is the fun part so that's mostly academic.

Strava has no clue about how un-aero I am, so Strava's power over a ride is a guess. But Strava knows pretty well how long it takes to go up a long, steep hill and going slowly means that aerodynamics is a very small part of it. So with real weights, and a long hill where Strava has good elevation data on, the power estimate will be very close IMO.