If I did not misunderstand your concern in the last paragraph, I would recommend you to check out the paper Fyhn et al.2004 where they actually show that you can predict the rats position from as little as 8 simultanous recorded grid cells. Furthermore, in the Hafting et al. paper (2005) it was even shown that the firing fields of neighboring cells was distributed (Figure 4). Together this suggest that a local ensemble of grid cells are able to code for any location.

PS! If you are interested in the details check out: Fyhn et al., 2007, Solstad et al., 2006, or Blair et al., 2007.