The ability itself being too wipey (due to reduced scatter) is a different story though. That probably is due to a lack of ample play testing; it takes a large sample size. Just look at this video that tightrope did in which he tested the ability:
Yeah, the example in the OP was rather extreme, I guess the results that tightrope got where actually what was intended.
Problem with this really is you need a lot of playtesting for the ability to realize that it can become excessively destructive. The things that have to happen at the same time:
1. The planes need to lock on to an infantry target.
2. The infantry target needs to retreat at the correct time.
3. The bombing run has to reach the target after it made it to base.
4. The base must be packed with units (note, there were also other units retreating into the bombs and the damaged Sexton)
5. The bombs have to hit said targets which alone is pretty RNG.
All of these circumstances have to be met to make the ability that destructive. Problem with beta tests is that people might have tried out the ability (probably being pretty critical of it to begin with), don't get "lucky" according to the list above and decide it is "meh" and not to use it again. So, yeah, these things are tricky to catch in the testing phase with a limited number of players.
The thing with loiters is that they are supposed to work as areal denial tools, so changing them to single passes wouldn't really have the intended effect.
Not sure if that would work, but is it possible to "untrack" an infantry unit when it retreats? Or make it impossible to track infantry units in the first place (only installations, vehicles and team weapons)?