The optimal amount of train delay is non-zero. You need high redundancy to avoid many causes - and their knock on effects - and redundancy is antithetical to efficiency.
It's immaterial though. They could instead face increased cost in taxes, decreased investment elsewhere, or good-old eminent domain for infrastructure expansion. The downstream effects of an infinite number of backup lines to guarantee service following an infinite number of track trespassers are not good for passenger satisfaction.
Regardless, my point is that the optimal amount of delay for profit and passenger satisfaction is not the same.
Ever since I’ve seen Japan’s trains drive perfectly as long as it’s not raining, on segments as short as 2 minutes apart, its become abundantly clear to me that delay is not an inherent property of any rail system.