Hybrid Bowden extruder? Anyone tried this?

I’m looking to do a really large print soon, and looking to cut down on print time by increasing my print speed as much as possible. One hurdle I see repeatedly is the weight of the X carriage limiting the acceleration. The typical ‘fix’ for this is a Bowden extruder. The problem I see though is about a 50/50 ‘I like by Bowden’ response. Lots of issues with bad prints, huge amounts of retraction necessary, jamming, etc. Without a Bowden extruder being a ‘for sure’ fix, I’m very hesitant to do the work building/installing one. Plus, given general physics, I can see a bunch of issues with a Bowden with my print.

So I had an idea - has anyone ever tried this?

What if the X axis held only the nozzle, just like a bowden, thus reducing the weight by a huge amount, but, above it, a second X carriage. This X carriage would shadow the lower X carriage movement, but carry all the weight of the extruder. It would use a separate X motor and belt, like how the TAZ has the dual Z motors, that way if it’s weight caused lag/issues, it should not effect the print quality of the lower head on it’s own motor.

In fact, if someone were feeling ambitious, the firmware could be modified to run the upper X axis off a completely separate motor controller (for those with single extruders, hijacking the second extruder output for it) with settings that would read the G code for the movement of the X axis, and mirror it with some ‘slack’ so that it could follow the lower axis, with different accel settings so that it’s movement could be smoother, thus causing less jerk.

Might that solve some bowden issues? I’ve never run a bowden but from my understanding, a lot of issue comes from the length/compression of the filament - it’s being pushed into the extruder from a good distance away, and the motion of the X axis will cause the filament to flex/move as well.

If the bowden extruder is just a few CM above the hot end, following it closely, but with less aggressive acceleration, it seems you should be able to get the weight reduction/speed increase of a bowden, while also reducing a lot of the issues inherent with the length of filament/distance between the bowden extruder and the hot end?

Has this been done/tried before? Does it sound like a possible option? If my theory is not completely absurd, and it hasn’t been attempted before, I might just give it a shot, for no other reason than to just learn something.