Yes, Ihave the same problem, and I'm trying to figure out where it comes from.
To the people saying it comes from the inaccuracy of the filament, I tend to think that this inaccuracy would be multiples order of magnitude lower than what we have AND not repetitive.
What we have is a very repetitive problem, so that's nothing we can't fix in the slicer.
Now, I wonder how to tackle that because it is not a constant problem accross the sizes of test elements I have printed so far. I came accross this problem while printing cylinders, but they are not good to investigate this since the slicer has to transform them in polyhedron anyway...
I'll try to investigate more