They will receive less BTC each halving, that's true but imagine that the price needs only to double during that time (around 4 years) and that's not very unrealistic that the price will be doubled in 4 years (1BTC = 14,000$ in 2023).
Right now, the hashpower is already quite high and we are in a very comfortable situation that 51% attacs are far from being performed against BTC. Even if the hashrate wouldn't be that high an attack would be a waste of money (The hashrate of BCash is currently 25 times lower than BTC just to show that).

So I can't imagine that anyone will do this, he'll lose a ton of money. Here's a good website where the costs of an 51% attack is calculated (estimated) and there are still many costs missing like power costs per day / setting up all the miners and much more: https://gobitcoin.io/tools/cost-51-attack/ (that's only the hardware cost)

And there will always be revenues from transaction fees. Right now that's up to 10% of the block rewards: https://btc.com/stats/fee

---

The main advantage of asics is that they can't be used anymore after the attack is done and they are just a worthless door stopper. Or at least their worth would be significant lower. ETH uses GPU (graphic cards) and you can still sell them later (until better ones are out), so an attacker would have a much lower loss compared to asics. This advantage is also a disadvantage for normal users because asics are almost worthless after they are outdated or broken and can't be sold anymore. Their only purpose is mining SHA-256 algorithm. Maybe thay can be recycled for precious metals. So both ETH and BTC PoW have advantages / disadvantages.
But for security reasons to protect the network against 51% attacks the BTC PoW is a better solution.