I think the only reason such people would prefer mining is if it’s difficult for them (legally, practically, etc.) to directly buy bitcoins in the area they live or at the scale they desire to invest.
Estimating how much it will cost you to mine X Bitcoin is not an easy thing to do, especially now that each block almost always has at least an extra Bitcoin worth of fees, with the highest having over 4, meaning the amount of Bitcoin you get per block can vary by approximately an extra 1/3 and this extra value is pretty random.
The power to run the hardware is also constant cost, not a one-time cost like buying a bunch of Bitcoin at market price, so while it's easy to say "Just buy when it's at X value and turn off your hardware", the miner has no way of knowing what that value is, and without knowing the future values of Bitcoin they have no idea if waiting and spending all their current savings on Bitcoin at a low price will actually net them more then just waiting it out and continuing to mine the whole time. It's a crapshoot, but the miners presumably know what they're getting into.
My point was more that it is probably acceptable to risk mining at a bit of a loss if you're confident the value is going to go up, you already invested in the hardware anyway, and you think there's a good chance you may make more then if you had just bought some at what you think will be the lowest price. That said, my example is probably more extreme then would be acceptable.
Miners have continuous liabilities in USD and so most only hold Bitcoins as long as as is necessary by the protocol (100 blocks). Further, expected fees are easy to estimate over medium-to-long time frames. I basically disagree with everything you've said here and below.
As soon as it's more profitable to just buy the bitcoin, the miners will turn off the machines. There's no point spending $1.01 to mine $1 of Bitcoin when you can just buy it at market rates.
It's not really that simple because mining Bitcoin is a lottery, and the amount of coins a block is worth is not a constant value due to being able to collect a variable amount transaction fees (Which are becoming much higher then they used too).
You can estimate approximately how much it will cost you per Bitcoin, and you can obviously view historical data on how much the mining has cost you, but there's no way to know going in that it's going to cost you $1.01 to mine $1 of Bitcoin, there's too many variables in play.
Furthermore if the EV is zero, you don't take the bet either because lowering your variance is a good thing.
The network already has a system in place to combat this issue - fees. So it can technically cost $1MM to mine 1 BTC while the price of BTC is only $8,000, as long as the transactions provide enough fees to offset the costs. In practice I think that's pretty unlikely to work out that well, but that is the intended system either way.
Edit: Reading the article again, it's not clear to me they're actually taking halving into account in their calculation, and I'm also not extremely confident they're not confusing bitcoins and blocks. There are no direct quotes detailing that part of the article, and I wouldn't really expect the person writing the article to know the difference.
And it also occurred to me that something else to take into account is that miners themselves are betting on the price of bitcoin going up. So just because it cost them, say, $200,000 to mine $100,000 worth of bitcoin, if the price of bitcoin eventually doubles from the point that they mined them then they would still break even, though not immediately. So it can make some sense to mine bitcoins at a bit of a loss. I don't think it would make sense to mine at anywhere near the losses they're describing though.