My suspicion is that the legal request is a lawyer's pro forma procedure written in the 1980s for collecting evidentiary information in a data processing environment: "Obtain a list of the tapes upon which the data resides". It's likely that they don't actually care what tapes are involved: their goal is to generate a list which contributes to an impressive compendium of factoids which fill lawyerly paper.
Given the size of TSM tables, my inclination would be to narrow down to a set of tapes, which you know by virtue of common storage pool and perhaps collocation by node. Then perform 'query content' on each volume, output to a file per volume, which you can subsequently grep for containing the files of interest. Query Content is tailored and efficient, greatly reducing time and impact. Richard Sims not a lawyer