I think the assumption of a 51% attack is not worth the effort is wrong. Eventually the price of Bitcoin will stabilize. Let’s say at 1 million USD then a diminishing block reward and the transaction fee need to compete with a 51% pool that can make arbitrary amount of profits. The transaction fees will have to pay for 51% of the network to keep everything secure.
Apart from other reasons why this won't happen. Let's reason if it's worth the effort to do a 51% attack.
So let's say are able to acquire the hardware to form this 51% attack
This would be quite an investment at the current hashrate, which is +/- 100M TH/s
A modern miner like Antminer S19 Pro costs $10k and runs at 110.0 TH/s
So you need 900.000 of these to get that amount of hashing power, which would cost you 900k * $10k = $9 billion. Of course you 'just need' 51% of the hash power, so let's say it's $4.5 billion.
You now can only turn back transactions if I'm correct. So you can pay for something in BTC and quickly fork the network to undo a transaction.
Once you did this, everybody will see this. Probably people will immediately stop accepting BTC now. And you cannot trick other people and have free payments.
Ergo: It's hard to imagine you win back your investment in mining equipment before people realize you cannot trust a BTC payment.
And I even left electricity out of the equation (small costs compared to mining investment)