This is a very interesting set of 3 ride files, and thank you for providing detailed descriptions of each so that I could interpret them properly.
The big puzzle is between your 5/6 ride, where you set your threshold power, and your 5/9 file, where you get significantly lower results.
I analyzed the threshold stretch on 5/6, and the comparable section of your 5/9 ride
I see the following:
1) Your PP looks to be properly calibrated. I say this because wind speed and slope readings are within range
2) Wind strength and direction varies considerably between these rides
3) On your 5/6 ride you had a slight headwind in the threshold interval. An aero coach analysis shows that PP measured 234 watts is very close to calculated 233 watts

- Screen Shot 2020-05-09 at 5.45.51 PM.png (185.92 KiB) Viewed 3894 times
On the 5/9 ride you had a pretty significant tail wind in the same section. Also, your bike speed was lower. You were not working as hard...

- Screen Shot 2020-05-09 at 5.43.42 PM.png (194.19 KiB) Viewed 3894 times
PP measured 163W; aero coach calculates 162W
In both cases, calculated and actual watts are extremely close!
The power data suggests that you did not work as hard on the 5/9 ride, and the sensor data (wind, slope and speed) support this conclusion.
The only anomaly I see is your HR reading--on the 5/9 ride your HR is higher, even though you aren't putting out as many watts. HOWEVER, the temperature on the two rides is considerably different; on your 5/9 ride it was a cold 7C, vs. a nice 18C on your 5/6 ride. Whether or not temperature accounts for the HR difference I do not know.
Be very mindful that how hard you work is going to be heavily influenced by how hard the wind is blowing.
I think your PP is properly calibrated, so I would do another ride without changing anything.