I've done a (very bad) sketch. The trick is in fixing shifter and servo together securly, and allowing the pull wires to go slack without becoming misplaced. (and having sufficient torque)
A trigger shifter set up like that would avoid the need for the electronics to understand more than "shift up" and "shift down", but as has been repeatedly said, electronics are easy.
I was thinking a traditional thumb shifter like we had in the 80s would be a lot simpler to interface mechanically. If the torque was sufficient you could probably even couple it directly to the servo arm. Intuitively, a non-indexed shifter should require less torque?
The answer is of course obvious when someone points it out to you. Standby for another shockingly bad sketch.
Worm gear on motor turns cable windy mechanism. When in place, remove power from motor - cable cannot back drive the motor. Attach a pot to the windy mechanism, and you've got position feedback. If cable pully part of windy mechanism is about 20mm diameter, then less than 1/2 a turn is needed for the full pull, which leaves lots of space for electric adjustment (2.5mm/gear on nine speed)
Will probably still need a gear on the motor before the worm gear, to get the torque necessary - so a modified servo (modified for continuous rotation) would do the job.
PS - have a great day tomorrow.