The ions are fairly mobile, and will adapt to the field of the less mobile charges within a few 100 ps of simulation.
The ion placement loops would have to be modified to exclude some more water molecules, based on their coordinate values.
The energy evaluation in the calling script uses PBC, so if the placement caused a major energetic problem, the config would not have been the lowest energy coordinate set chosen.
If you think the starting config may be a problem, one option is to change the random seed and run the ion addition again. I often do 5-10 runs with different random seeds, and compare the final coordinate sets produced. If you have the computer power available, it's also not a bad idea to run simulations in triplicate, each with a different initial atom packing.