Firstly, auto speedbrake is not a function of software. It is actually a quite simple circuit where a bi-directional actuator, actuated by air ground logic and thrust lever position is operated. A secondary circuit using thrust reverser lever position will also deploy the speedbrakes if the T/R is deployed.
As for a modification to enable them to be stowed on application of t/o thrust. I don't think it would be too expensive. Fag pack calculations lead me to feel a second actuator would be required (as B73567AMT says the current actuator needs to be extended in order to auto retract) which would be operated by a microswitch on the throttle switch pack (it is possible a spare already exists I don't have the WDM to hand) It would need to be bolted to structure and could drive through the 'no back' clutch already used in the existing installation (to allow over ride by crew in the event of a runaway/UCM). So total cost 1 actuator, 1 (possibly 2) microswitches, about 100' of wire (22awg for control and 18awg for power) and a few hours on the drawing board for a designer and a few hours for the STC or type approval from the FAA/CAA. It's not a complicated circuit as no air/ground interface is required, full thrust apeedbrakes down and that's it. It wouldn't interfere with the ground autodeployment function as this too will stow the speed brakes if full thrust is commanded and the a/c is on the ground.
As for why it was not incorporated in the first place, remember the 757/767 was first conceived in the 1970's and auto deployment/stowage of ground speedbrakes was at the time quite an innovation. they didn't want to get ahead of themselves by thinking about air operations too.