Arguably you should preserve causality in this question. Indeed, "foot pushes door, so door pushes foot" could, in the context of N3L, be written the other way around and still make sense mathematically, however we must remember that the time ordering of events is still important.
Consider the initial rest frame of the door and foot (i.e. at rest with respect to the Earth's surface). The foot is definitely accelerated first by the muscles in the body - you cannot say that the door is accelerated. Relativistic symmetry is broken here - as a "little person on the end of the foot", you would know that it was you who was accelerated towards the door, and not the door that was acclerated towards you. So you cannot say that "the door is travelling towards the foot".
So, it stands to reason that - in the initial rest frame - the foot provides the "action" force on the door, since without the foot's prior motion, the door would not be able to provide the "reaction" force.
Though, of course, at the time of contact between foot and door, mathematically the forces are equal and opposite, so causality is irrelevant.