Hi. I have just reuploaded version 3.22.
This is certainly not the ideal state for this ancient package, but I believe it's slightly better than keeping an important regression forever. If upstream decides to maintain procmail again (which means addressing regressions, not merely making a new release), it could make sense to consider the switch again. I was about to write all this in the Debian changelog, but this place is probably more appropriate. Thanks a lot.