Can you explain for a layperson why that is?
I think the odds are MUCH shorter than that.
It's going to depend on a lot of factors, but the manufacturer likely has a target capacity that they hit fairly closely; If you are selling 300mAh batteries, then you don't want to have many leaving your factory with an actual capacity of 350mAh, because that implies a wasteful production process; And you don't want any below the specification, if you're ethical and/or concerned about being sued or fined for selling products not fit for purpose.
The other major variable is the current draw of the smoke detectors, which for similar units will again be unlikely to vary very much.
If the variation in battery capacity is 5mAh, and the current draw is exactly 1mA (numbers
ex ano for illustrative purposes only), then the odds of failure within plus or minus 2.5 seconds are very crudely one in the number of 5 second intervals in 5 hours, or 1:3600
However, this assumes that the distribution of actual capacities is random. More plausible is that they are normally distributed, with far more batteries having capacity close to the central value than have capacity at the extremes of the distribution (or even outside it).
This could easily bring the odds down to one in a few hundred - and that's for each pair of smoke alarms, on each occasion that the batteries are simultaneously replaced. Replace the batteries simultaneously every six months for a decade on a half dozen smoke alarms, and the odds of two simultaneously running out could be very high indeed - certainly high enough to be fairly unremarkable if it happens.
But the number of variables is very high, so any odds you calculate are going to be very woolly indeed.