The funds aren't locked up and can be freed at any time.
It's true that the funds aren't really locked up but they are removed from the exchanges and out of temporary circulation.
If there was a 0.2% trading fee, you would need to hold the position for a certain period to make the yield worthwhile, so there is some lock up incentive.
A yield harvester would require twice the collateral, 100% on each side? And pay fees on long and short side? Thereby halving the effective yield and doubling the fees, thus incentivizing a much longer holding time to make it worthwhile. (I don't know if that's true, hence all the question marks.)
The problem is you never know if someone is really short
By that do you mean, we don't know if they have an equivalent long position that cancels it out? If so I think that's still a benefit as that requires more BTS which is also being taken off the centralized exchanges, out of temporary circulation and is helping add to the BitAsset CAP which is a big marketing positive too imo.
Edit: Also because of the forced settlement function, yield harvesters are at risk too, they would have to at least pay trading fees again to re-open their short position if forced settled or add more than 100% collateral which wouldn't receive yield. (If it was possible to only pay the first 100% of collateral.)
Admittedly there is a lot of things I don't understand about this, I'm just trying to think of solutions that would help create a tighter peg.
Edit2: If you implement MAKER instead, I think you may find that when the incentive runs out in a few years but the same market conditions persist, then shorts will again be reluctant to short close to the peg and we'll be in a situation where we can't use trading fees to incentivize shorts because they're going to pay Makers for past services rendered.
This is also cheaper because we only need to incentivize shorts when BTS short to medium term price expectations are neutral to negative.