The question is of course: which incentive could work here? You need a target to attack if you launch an 51%. And this target has to be valuable more than the resources spent on the 51% attack.

Just for fun, I googled "sell testnet bitcoin" and there were indeed some results. But the first Google listing, sidesift.ai, has the trading pair "paused". Then there is a site "Buy Testnet", which seems to be an one-way exchange of some miner, selling each tBTC at 0.0000045 BTC. However, the attacker needs to have access to the sell side, i.e. he has to be the one selling or depositing the tBTC, otherwise the attack would not work. On "Buy Testnet" this isn't the case, it seems.

According to CoinPaprika there is an exchange called "AltQuick.com" which allows to buy and sell tBTC. Indeed this one would allow the attack, it even accepts coins of two different testnet versions. Price is at 0.00000272, but orderbooks are very thin. So let's say you 51% attack AltQuick.com after a deposit of 10.000 tBTC and expect to get 0.0027 Bitcoins, you only would get about half of them. That would be about $110-120. And a much higher deposit doesn't make sense because there are only 0.004 BTC offered for this pair. So in the best case you would get $300 with the attack.

Would really $300 be worth the hassle? Perhaps it would be worth it as a call to AltQuick.com to delist that pair.