There might be a few reasons for that.
First is that if you haven't imported all the items into the track yet, then the shader count is still low, so that is fine. If this is the case and you add more and more shaders, then having reduced 15 to 2 shaders will benefit you later!
If that isn't the case, then it could be that the 15 road parts are often not all seen together in the first place. Perhaps you only have 3 or 4 within your usual draw distance at the most.
There is also the downside that you can't now give each road texture a different material property, like more/less grip etc... so sometimes for these things it's ok to run more textures/shaders as there is no avoiding it if you want many materials!
Obviously managing all this is time consuming and can be quite complex. You really need to get an overview of what textures/shaders are used where.
Ie, atlasing a shader for a building at one end of the track with a building on the other side of the track when you never see the other will be pointless. Just LOD the other off after it goes out of sight and you've saved yourself some work.
Making a track for us is as much about considering these things as much as anything. We don't have scripts/workflows that can do all this for us... so it's a job we have to manually work through.
Infact, originally, these track resource might have been heavily optimised with files within the rFactor engine... but they won't move over to Racer easily if at all. They are totally different graphics engines an era apart!
Optimisation is important, don't lose faith in it yet.
See those 10 or so saved shader calls as extra potential later to add more details, scattered grass objects, add more bump/spec maps etc etc...
Every bit of optimisation lets you invest elsewhere to make stuff look even better
Just as an example I have. On my course I have about 300 cones. They are all individual DOF so they can all be knocked over as movables.
Downside is that is about 300 render calls!
So I simply added very close LOD drops, so the models go away by about 150m for the small cones.
Now suddenly I lose about 250 render calls. I still get about 50 or so as 50 cones are in view at any given time, but the FPS benefit is really clear!
Just work at it, I swear it'll be worth it in the end... just try look for the worst culprits for hogging shaders for now.
In rFactor tracks the number one problems are trees and track side items.
Again, on Roggel atlasing the trees got me about 25-30% more FPS, so it's well worth doing it in cases where there are clear cases of many textures all being visible at once, but in separate shaders.
Look out for those types of things and they will get you the biggest gain
Thanks
Dave