Trackers are attractive because the maximize the production from a fixed number of panels. It is only when you actually do the math that you realize that maximizing the production from limited number of panels isn't the best optimization goal.
As others have said, trackers improve production but cost more than the benefit you get. For a given fixed number of dollars, you are better off buying more panels than using trackers to get more production out of fewer panels.
This is the same sort of math that plays out when you realize that you are better off having a DC:AC ratio > 1; selecting an inverter that can squeeze every last drop of power out of an array gets you less kWh/$ then building a system that clips at peak times.
In locations where energy is worth more at particular times of day, it pays to have panels that aren't pointed for maximum production, but rather maximum production at those desirable times.
Land is also a big cost for PV, so arrays located to use cheap land rather than optimal land for solar production can give you better kWh/$.
I believe that single axis trackers are economically attractive in some agrivoltaic applications, where you need to optimize plant growth and electric production; in this case the trackers might actually _reduce_ PV production.
IMHO the 'solar roadways' idea was always horrible, but I do like the idea of solar canopies over roads. Lots of land, easy transport for maintenance, and maybe it would keep the snow off the roads.