The author does, explicitly, consider benign intentions. But beyond serving as a read receipt it also includes location data and timestamps for each 'read' which is much harder to consider as due to a benign intention (from the perspective of the recipients).

Other email clients support 'read receipts' without using tracking pixels. It's not impossible that Superhuman chose to use tracking pixels for a good reason, but I can't think of one and I haven't read of one either.