No, the difference is not the tail energy. Tail energy should be -140, but the difference is 140.
I find if I change the command:
pair_style lj/cut/coul/long 9.0
pair_style lj/coul cut long 9.0
then it is consistent, and the total potential pe is equal to evdwl + etail + …
Values of other terms are not changes. Only the evdwl changed, which is 140 less now.
Does ‘pair_style lj/cut/coul/long’ have some problems?