Because the technique of measuring running economy (i.e. by measuring respiratory gases) assumes that the energy consumed is aerobic. If you pick too fast a pace to measure running economy, the anaerobic contribution will get too big and distort the results. In this case, they determined that increasing the pace to 17kph would have pushed some of the runners above 90%VO2max, which would have been too high for accurate measurement.
It's true that the p-value was 0.07 (i.e. not smaller than 0.05) at the fastest pace, though the rearfoot group's average energy consumption was still 5.0% lower than the midfoot group. It would definitely be interesting to see what happened at higher speed, using a different measurement technique.
I have no idea what you're trying to say here. "Longer contact time and shorter flight time" doesn't mean they're slower -- they were running at exactly the same pace when those measurements were taken. And, in case you've forgotten, they had equivalent personal best times (~70min for HM) on average. That just reflects the fact that if you land on your heel, you spend a higher percentage of your stride cycle in contact with the ground. It's neither inherently good nor inherently bad.
That's incorrect. Since previous studies have grouped forefoot and midfoot strikers together, this study did the same for consistency: "Therefore, in this study runners who appeared to land on the ball of the foot first (i.e. forefoot) or who landed with the heel and ball of the foot simultaneously (i.e. midfoot) were grouped together as midfoot strikers"
You're a strange little man.