I'm not sure about the US, but in Australia pretty much all traffic signals in metro areas are all linked and controlled over fibre networks. You need single mode fibre given the distances between cabinets etc. (hundreds of metres to kilometres).

And if you're doing that, may as well just use single mode internally too because the transceivers and patch cables aren't really much different in price and you can get better volume discounts.

Without seeing the ferrule to ferrule contact it's hard to say if the termination is really dirty. Yes the jacket is dirty. What they have done that is a cardinal sin is running the fiber between equipment through the center of a rack rather than around or above.

I hope whoever installed that has a good cletop, fiber scope, set of ferrule sleeve cleaning swabs, push click cleaning tool.

Also hard to say if it's dirty or bad without seeing the back/internal side of that patch panel. Knowing city governments it could be anything from a dreadful mechanical splice to a perfectly done core alignment fusion splicer with 0.01 loss.

Could it be connections to other nearby signals for synchronizing the lights?

Also possibly a connection to a central traffic control center that can override signals during emergencies or special events that generate abnormal traffic flows.

I mentioned SCOOT elsewhere in this discussion. With such systems, overriding is (I was told) the norm. The isolated controller on fixed timing is the fallback case.