Yep, I totally agree with the idea that extrapolating from static tests will be near useless. At least not without a lot of real world data to use for coming up with an algorithm.
I gave that example of a same weight and same power input sailplane vs racer above. That's an extreme of course. But I've been on the receiving end of a couple of real world examples of this myself. A four stroke glow engine model that was marginally powered which was fun to fly with a given size and brand of prop which became doggy and noticeably underpowered when flown with the same size prop by another maker. In that case even the shape of the prop came into play. I've also seen the same thing with the early electric models that used heavy battery packs and brushed motors where an old timer design's climb performance altered with minor changes in the prop.
Another point to ponder.... Years back a flying buddy bought out the old Y&O propeller line. He always found that these worked better than many options for his style of model flying. To try to find the cause he made up a pitch testing jig and spreadsheet to analyze the real pitch distribution from root to tip. He found that there isn't a prop on the market that carries the stated pitch value from hub to tip. And many didn't really match at more than a very small portion of the blade. So start by assuming that no prop from any maker is actually the pitch it says it is. You'll want to mimic this measuring yourself.