Hi Charlie,
This isn't something specific to tbss_sym, but related to any analysis that compares both hemispheres. During the process of symmetrisation, the structures have their positions shifted slightly. Because we should always identify structures using the anatomical relationships between them (this implies visual inspection of the results), these shifts are not at all problem. You certainly have the mean FA map after symmetrisation, which could be used as a reference and, ideally and much better than that, the average structural (T1) scans registered to the same (symmetric) space, so overlay the results on them for anatomical reference should not be an issue.