As with all things related to design, it’s all down to a balance between multiple factors and requirements, with the laws of diminishing returns coming into play.
There was a famous experiment where someone made a bent barrel, that was nevertheless able to shoot accurately, proving you only need the last few inches of the barrel to be true to shoot straight with it.
Usually people want longer barrels for more powder burn, so they get more FPS. That translates to less time to target, resulting in less bullet drop and external influences, like wind, you need to take into account.
Barrel oscillation is a known problem, which is why gunmarkers typically compensate by making a barrel heavier the longer it gets (they could alternatively use better materials to keep weight down, but that carries a monetary cost).
You also need to consider that what is acceptable oscillation for a bolt action sniper rifle with a free floating barrel may well not be for an automatic military weapon without the free float barrel.
In terms of delays to subsonic, for a 50bmg, the bullet will only drop below the sound barrier after 2.6k yards per this calculation. So is it really worth the costs to add a longer (and heavier barrel) for the sake of maybe pushing that back by a few hundred yards more?
Also, I would not take the line of reasoning that ‘because people buy it it must work’ as all that reliable, as people buy some pretty stupid stuff to bling out their guns a lot of the time.
What some shooters want for fringe needs (like competition shooting, or tacticool) is not necessarily a good reason for general military issue.