A tiny bugfix for the reset flag, and a brief explanation of the changes

[Syntax:]

fix ID group-ID ipi address port \[unix\] :pre

fix ID group-ID ipi address port \[unix\] \[reset\] :pre

ID, group-ID are documented in "fix"_fix.html command

ipi = style name of this fix command

address = internet address (FQDN or IP), or UNIX socket name

port = port number (ignored for UNIX sockets)

optional keyword = {unix}, if present uses a unix socket :ul

optional keyword = {unix}, if present uses a unix socket

optional keyword = {reset}, if present reset electrostatics at each call :ul

[Examples:]

fix 1 all ipi my.server.com 12345

fix 1 all ipi mysocket 666 unix

fix 1 all ipi mysocket 666 unix reset

[Description:]

initial configuration is ignored, as it will be substituted with the

coordinates received from i-PI before forces are ever evaluated.

A note of caution when using potentials that contain long-range 

electrostatics, or that contain parameters that depend on box size:

all of these options will be initialized based on the cell size in the

LAMMPS-side initial configuration and kept constant during the run. 

This is required to e.g. obtain reproducible and conserved forces. 

If the cell varies too wildly, it may be advisable to reinitialize 

these interactions at each call. This behavior can be requested by 

setting the {reset} switch. 

[Restart, fix_modify, output, run start/stop, minimize info:]

There is no restart information associated with this fix, since all

  inet   = ((narg > 5) && (strcmp(arg[5],"unix") == 0) ) ? 0 : 1;

  master = (comm->me==0) ? 1 : 0;

  // check if forces should be reinitialized and set flag

  reset_flag = ((narg > 6) || (strcmp(arg[5],"reset") == 0) ) ? 1 : 0;

  reset_flag = ((narg > 6 && (strcmp(arg[5],"reset") == 0 )) || ((narg > 5) && (strcmp(arg[5],"reset") == 0)) ) ? 1 : 0;

  hasdata = bsize = 0;