I think the way to go is run a first pre-pacing simulation of 10000ms without writing down the igb (or just writing a little amount of points so that you are not storing a huge amount of data). Save the state variables at the end of the prepacing simulation:
'-write_statef', writestatef,
'-num_tsav', 1,
'-tsav[0]', tend_state,
And run a 1000ms simulaiton
'-start_statef', startstatef,
EDITED: you can use the 'spacedt' option as the duration of your simulation to reduce the igb file of the pre-pacing to just one point.
cheers,
Teo