'master list distance cutoff' changes strangely after restarting the simulation

Dear lammps user:

I want to compute the gyration of radius of a single PE chain using airebo
potential. After a simulation, I continued the simulation by a read_restart
command. Then a strange thing occured that the 'master list distance
cutoff' changes from 12.2 to 8 in the log file and the gyration of radius
also changed a lot. I did not change the pair cutoff and the neighbor

​thanks for being so diligent and noticing this and also *many* thanks for
providing a suitable complete(!) input deck to reproduce this.
we will have to take a closer look at what is causing this. this is
something that should not happen.

there is a very simple workaround: you should move the "pair_style" command
_after_ the read_restart command. that will initialize the cutoffs