I see where you went here, and it makes sense in a certain way, but its a bit overthought. It would be great to have a magic lens to see how long a fight is going to go and not inflict any overkill damage, saving buffs and GWM hit-impairing uses. But in reality, up until the end of the fight its not really clear whether we can "hold back" on damage to avoid overkill. In the end, the value of dropping creatures 1 round earlier is much higher than any other damage, and that's why we pursue causing damage.There's an error in the basic premise that gets back to overkill as a limiter. Potential DPR is never realized unless it actually reduces the number of rounds ending combat or killing / defeating a creature.
It doesn't matter if a person does a theoretical ~20 DPR on an ogre via -5/+10 needing a 5 normally vs ~12 without because the fights don't have enough actual attacks going on for the law of large numbers to come into play and theoretical DPR is trumped by real DPR. Real DPR is capped by monster hp and overkill is lost.
IE. If it takes the party 3 rounds to drop 3 ogres the party DPR is 59 regardless of probability. In order for the feat to increase DPR it would need to decrease the number of rounds of combat regardless of an increase weapon damage even if the law of large numbers had an opportunity to take place. If the fighter adds 8 damage (needing around a 5 to hit per above) each round that 24 extra damage isn't enough that it encompasses one Ogre's worth of hit points and the party is likely still going to need 3 rounds to drop 3 Ogres for no real DPR increase.
Theoretical DPR increase from the feat is situational and largely an illusion based on an average that doesn't have a large enough sampling to be realized.
So while it is true that damage could be wasted if a fight ends and the party has swings left, this phenomena is not particular to GWM damage. In a sense, you could therefore discount for instance 10% of total party fight damage as wasted in a typical 5 round fight (half of a round, and assuming 20% of your total fight damage was inflicted that round). But "end of battle wasted damage" and its resulting drop of real dpr vs actual dpr is irrelevant to our discussion here because it affects BOTH sides of the equation/comparison (+2 STR vs GWM, or "take your pick combat feat" vs GWM).