If the tang was designed for ejecting the taper and not as an anti-rotation device, why would it have two flats? You could eject it with just a reduced diameter stub on the end.
The tang is so that when you use a sleeve and drive the separating feather /key in to the sleeve to separate the drill you don't mushroom the softer dead end of the drill & cause it to jam in the sleeve .
Remember , morse tapers are a system of using friction at /in precision faces to drive the drill , they are not specific to one model of lathe .
So an 8 mm MT2 could be sleeved up to fit into an MT4 or MT5.
Thinking about it a bit more & still thinking about what I've already said perhaps by only taking the taper so far down and then making the slot for the feather at the end of it just happens that the tang goes into the slot area as the slot may well be at the end of the length /depth of the taper quite by accident rather than by design .
I've not found any tangs that fit snugly in the taper at 90 degrees to the feather slot they all stop well before that and you have to rotate the drill to get it to go in any further.
Could this be because the feather slot has been hot punched in the blank sleeve to make broaching /milling it out easier ?
I do however had a couple of morse taper drills that came with the lathe that have been badly ground or turned down on the tang end .
I assumed it was done because of the mushrooming of the tang happening when being brayed out using the wrong face of the feather or an incorrect sort of feather such as one made from ground down tool steel . or When someone has walloped the drill in a slipping/ scored /burred /dirty taper to make it hold and then couldn't easily get it undone again .