[A sample of] new breakers are tested at the factory for 10,000 tries.
You could test your breakers 100 times and still not be sure they will work on the 101st try.
http://en.wikipedia.org/wiki/Sunrise_problem
BTW, consecutive successes with no failure is "right censored data", so to do the reliability calcs I guess they assume it would have failed on the 10,000 plus 1 try.
But, how often will a house breaker be called upon to trip?
This is the "mission length"; how reliable is a breaker that fails in less than 10,000 attempts when it only needs to work on 1 or 10 or 30 overloads?