Here, we see an example of the evaluation code, which filters out those tracks which come from our embedded particles. It has found three reconstructed particles for one of our embedded pions. (The other five embedded pions only lead to one reconstructed track.) Clearly, this is an issue for HBT. The acceptance correction of course depends strongly on the cuts applied to the data.
The standard cuts on frac_seen and nhits would do more than enough to make this example not a problem at all (the 74-hit track would survive, while the 12-hit and 7-hit tracks would not). But it is interesting to see that the tracker (jtf in this case) can mess up even in such an apparently simple case.