I like the one comment lamenting people saying, "It can't be that difficult to implement." If you're not the person who is in the code and doing it, you don't know. You can't say what is hard and what is not hard. You don't know what the code currently does, what systems it makes available to support the change. Perhaps the feature itself is small, but getting things in place to support the change is not.
It may not be easy, but there are certainly several add-on products that support this feature and they have no where near the same level of resources and know-how that the Microsoft team in charge of the taskbar has. So it seems unlikely that "hard to implement" was a primary reason for not implementing this sooner.
But that certainly ties into the cost-benefit analysis. For the small percentage of users that would want such a feature, they are probably also the users most capable of finding those 3rd party tools. So the percentage of people for whom implementing this feature natively would actually benefit is incredibly small compared to the cost of implementing it, even if the cost isn't that high.
I deal with this with my client all the time. He thinks adding a column to a report is "easy" because "it's just one thing". He thinks changing a constant factor in an equation is "hard" because it dramatically changes the output.
It's the complete opposite. Maybe the data he wants in the column isn't being tracked. Maybe it's not even very well defined. But he is like all users, all he knows is the output, there is no intrinsic knowledge of the input or the process.
