It doesn't matter how large your detector is if the signal of interest is weaker than the natural noise. There are methods to find the signal in the noise but you need to know more - much more - about the signal. This is studied in great detail in the Information Theory.
So if the nuke doesn't radiate much more than a slab of granite that a contractor is carrying to install as a kitchen counter then there isn't much of a chance to tell the difference between the two. More likely you won't be able to detect either of those sources. To be successful you'd need to cover the entire country with detectors, and I'm sure Pakistani military will not be too happy about that.
It's also possible that the trucks that Pakistan uses to carry nukes have internal shielding. A common container can carry a lot of cargo - 20 to 25 tons.
Containers are also all alike, sturdy, and sealed - those are desired qualities for a covert transportation of stuff. A container can be quickly moved from a truck to a rail car, a ship, or it can be stacked so deep that nobody will be able to dig it out without some major time with the crane - not something that can be easily done by thieves.
Well, trying not to say too much, but in the 80s my brother worked on Bayesian probability applications in a very similar context: more or less continuously updating probabilities that the incoming is what you think it is; whether the incoming is a particle or something else.