You would be hard-pressed to make something more safe than a pneumatic solution. All you need is a safety valve that opens at a specific pressure to limit the amount of force. And that takes care of all probes. No probe that will press harder since each cylinder would have the same area.

Anything with motors, gears, ... are likely to be much too powerful so relying on software/sensors to stop when reaching the allowed force. Even friction couplings are problematic when you have 100 that all must be just as reliable and safe.

There are so low forces and no problem with air leakage so each cylinder can be even home-made by a rod with groves for o-rings and inserted into suitable-sized tubes.

A trivially cheap micro-PC capable of running standard Linux or Windows applications would be more than capable of handling quite cheap web cameras in case image-recognition is used for position detection. Etching lines into the rods for photo-transistors to count pulses should also be doable at a reasonable cost. Having a rubber wheel pressed to each rod and a rotary encoder would require quite a lot of rotary encoders. But should still be doable with that budget.

I don't see the hardware costs as the big issue. A bigger issue is to consider how to make the device practical to use, practical to store and practical to maintain. For example if a cluster of rods can be spread out like a fan when the device should be used, but allowing a much smaller storage size. 100 cylinders would mean 100 air tubes - not so fun to spread out 100 tubes from a central distribution cap like the ignition wires. Nicer if the air tube jumps from cylinder to cylinder.

This is mostly a mechanical project with potential for some fun programming depending on what sensor solution that is selected.