The program estimates an upper bound for the number of critical points (nr. of atoms times a hardcoded maximum number of CP per atom). When the actual number is more, the remaining points are ignored. So it is not really harmful, but some points will be missing.
Currently there is no input option to extend this array.
The maximum number of critical point per atom is now hardcoded to 256, rather a lot. It might be interesting to know the job that caused the problem.