I think you have something backward... if a short seller sells bitUSD at the feed price and buys it above the feed price they have lost money, not made money.
Ops, I confused that...
If bitUSD is priced above the feed on the internal market, it motivates short sellers to short on the assumption the exchange rate will eventually go back down so they will be able to cover later for less. But it is not specifically an immediate arbitrage situation in and of itself.
Is that enough to create the necessary BitUSD if there is a big demand for it? Why should the price of bitusd go down? It's not a clear decision for potential shorters to short on that assumption even if the demand for bitusd is big signaled by the high price of bitusd.
No other mechanism is needed to make the bitUSD price go back down simply because (above parity) there's no limit to the supply.
There is a limiting factor on the supply, being the market's appraisal of the value of BTS (BTS market cap). Once nearly all the BTS are in the collateral pool, there is no scope left for traders to create more bitAsset. BTS are traded in a free market, so there is no way to guarantee that the BTS market cap will remain far above 3x bitAssets. In such a situation a bitAsset premium could be sustained indefinitely. There is no mechanism in this situation that can force the bitUSD price down, except possible that potential bitUSD buyers realise that bitUSD is suddenly more risky.
There's no practical limit to the supply of bitAssets in part because there's no specific limit on the marketcap of BTS. The marketcap of BTS will be driven upward long before "all the BTS are in the collateral pool" and people will cover and re-short to free collateral.
While there's no theoretical limit, practically you can't create any more bitAssets than the BTS market is prepared to back. The market cap of BTS is determined in a free market, based on the market's appraisal of the prospects for future returns on it. Like any other asset market, it is forward-looking not backward-looking. On any day the market can reappraise the value of BTS to be significantly higher, or significantly lower. Therefore whatever people may believe about price pressures from bitAsset demand, any such pressures,
even if they exist, become completely irrelevant in the next moment of time when the market is free to reset its trading price for BTS.
So the defining limit on the size of everything is how much value the market is willing to place on BTS. The market cap and growth of bitAssets are certainly factors in this valuation, but they cannot force the BTS market by any means to being valued at more than 3x bitAssets (or any number at all) nor to growing alongside the growth in bitAssets. That's indeed why a black-swan event is possible rather than impossible.
While one might argue that if BTS rises it allows shorts to roll and free collateral for further supply, I could equally argue that BTS declines would lead to shorts losing collateral and reduce further supply. Both these statements are true. Which way it goes will depend on what the market ultimately is willing to price BTS at.
BTW your above explanation for why shortselling isn't bullish makes no sense at all. You say if someone sells some stake but uses remaining stake to short bitUSD it isn't bullish on the direction of BTS vs the bitassets as I stated in my paper. But no one would do that unless they thought BTS would go up in value relative to bitUSD so it's still bullish. This just describes someone who is bullish but also needs money for something else; it also isn't "arbitrage".
My original point was that a trader taking a short position may not be bullishly motivated in taking it, if they are hedging their position as one leg of a trade solely to take advantage of a bitAsset premium. The fact as to whether they hold any stake beforehand is irrelevant to their motivation behind the new trade - they have not adopted a more bullish stance. They are not seeking to increase BTS exposure, or having any net impact on the BTS price. Its further possible they may only own some BTS as inventory for this sole purpose rather than actually having a bullish outlook. I am still agreeing with you that the (naked) short position is a "bullish position" - but that's a quality of the instrument not necessarily the trader. (You're correct its not an arbitrage, but its the least-risk implementation to take advantage of the bitAsset premium).
Having said all that, this particular debate possibly seems out of proportion with your original intent.