Future Middle Imperatives in -ntor

On pg. 424 I hesitatingly try to explain the creation of distinctive future medio-passive imperative forms in -tor and -ntor as the result of the following analogy:

agit : sequitur :: agitō : X, X = *sequitōur > sequitōr

But Ben Fortson (per litteras electronicas) notes:

The passive future imperatives in -(n)tor < -(n)tōr in Latin you give with a question mark as coming from *-tōur with -ur from the non-imperative. But the Romans wouldn't have been in a position to abstract -ur as an ending until pretty late, after the change of -or > -ur in the 3rd cent. So wouldn't it be easier to say that -(n)tōr is simply -ntō + -or? Or, alternatively, what if what they really did was to abstract -r and not -or, as must have been the case to explain 1sg. passives where -r replaced -m, and which also presumably suggested itself for the 1st persons in -o(:)r where it looks synchronically like a simple -r has been tacked on to the active -ō. Then -(n)tōr would just be -(n)tō + -r. That would be preferable if you want to insist that the forms in

-(n)tōr weren't created until after the change of -or to -ur because of the survival of "active" utito etc. into the time of Cato and Plautus. (But that doesn't mean the -tor forms were only created that late; the two could have coexisted for a time, of course.)

I think that Ben is correct that working with -ur is too late. So maybe:

agit : sequit-or : agit-ō, : X, X = sequit-ō-or > sequitōr

But a possible problem here might be that if we push back the innovation to before the change of -or to -ur we might also be pushing it back before the loss of final -d after a long vowel, which also happened in the 3rd century BCE (Weiss 2009:155).

So probably the second solution outlined by Ben is preferable, viz. the simple addition of passive -r to the active form. As Ben notes, -r was extractable from 1st sg. act. -ō vs. 1st sg mid. -ō-r, and in fact -r was interpreted as the distinctive medio-passive marker as 1st pl. mid. -mur vs. 1st pl act. -mus and the alternate passive infinitive -ier (Weiss 2009:446) show.