I agree this (automated abuse detection lagged realtime, flagged some past behavior) is likely. I also suspect whatever else may be the case, the lack of detailed explanation is probably because whoever is fielding support simply doesn’t know (and may not be able to know) what was flagged.