My estimation is that a large portion of public transport users only use it for part of their transport needs thus making the fixed costs of car ownership a sunk cost.
Therefore, the variables costs; gas (or electricity) and maintenance make the most sense to use in living cost estimations. Even at the extreme end where you can live in the middle of a city and use public transport for mostly everything, you'd still need a car or a friend with one to go skiing.
We could have a debate about including it in a per mile rate, rather than splitting it out, but ultimately they are costs car owners have to pay, and non car owners don't.
A sunk cost is still a cost. A cost you have to pay is still a cost.
I think we could say it switches from a necessity to a leisure expense, and that city dwelling households can get most of what they want from a single car vs. several.
Clearly if a household moves from several cars to one car, fixed costs will decrease, which seems like an argument to measure fixed costs.
I'd agree with your point about vehicles potentially being leisure expenses, you could perhaps go further and start dividing up those fixed costs between leisure use, and business use. I'd even accept the argument that if you need a car, your leisure transport should be free from the fixed costs, as they have to be borne anyway, but they do have to be accounted for somewhere.
I'd be curious how many miles the grandparent has on the car though. My per-mile costs are also sky-high (because it's 10 years old and only has about 40k miles on it - I think it works out to $0.75/mile or so), but the flip side is that if this drives like a normal Honda it should be good for another 150k miles and, if I don't get into an accident or change my driving habits, I will never need to purchase another car.