a lot of people never did understood what TESTool actually did (or decipher it's reporting), and ghostwheel never really explained it fully either. TESTool is great, but it requires a modder to experiment to learn exactly what was affected. a combination of both of these is why TESTool had a bad rep -- modders using it didn't know what the report was telling them, and the end users of mods were using options that could cause issues (like 'JUST FIX IT!').
the most common dirt that TESTool corrects is removing object data from modders that keep hitting 'save' on references when tweaking stats like scale, ownership, etc, instead of just 'x'ing out. this is prevalent in a lot of mods, including GDR. using TESTool, many modders took the entries from the report of deleting those objects as deleting references in the world that they placed.
could TESTool break a mod? in rare circumstances i suppose. all of the mods listed "do not clean" i did clean with no ill effect. afterall, none of those mods ever explained why those dirt entries were essential. nor did those modders ever experiment with TESTool to figure out what was happening.