>>9844833You don't need Borel-Cantelli, just the -additivity of null sets. Given any such , any finite induced subgraph and any extra node , the probability that the induced subgraph on is in a given isomorphism class is some expression of the form , which is positive. So the probability that you get a given class for is one.
Now given any two such graphs you can use a 'back and forth argument to construct larger and larger 'partial isomorphisms' between the two. At each step you have a 'probability one' chance of finding the next , or in other words probability zero of failure. Therefore by -additivity there is probability zero of failure at step. In other words, with probability 1.