The best mechanism for that is old fashioned investigative work, e.g. in the North Carolina 9th district election.
The worry is that, if everything works this way, some of the indicators used to discover this sort of thing are diluted. E.g. the first indicators in the NC 9th case were that one county in this district had a lot more absentee ballots than elsewhere, and the absentee ballots favored one candidate by an implausible margin. If all ballots are absentee, the former measure becomes useless and the latter becomes a much weaker signal.
The worry is that, if everything works this way, some of the indicators used to discover this sort of thing are diluted. E.g. the first indicators in the NC 9th case were that one county in this district had a lot more absentee ballots than elsewhere, and the absentee ballots favored one candidate by an implausible margin. If all ballots are absentee, the former measure becomes useless and the latter becomes a much weaker signal.