We are not reading the correct loop restoration coefficients for a superblock

At the moment, we're assuming exactly one LR unit per superblock, which isn't always true.