I think one reason for that is the OO attitude you describe, where writing a module means a separate project done in isolation, hoping that someday maybe it just might be useful for something else, so you spend a lot of extra time trying to make it generically accessible.

Indeed. Though the problem is not confined to OO modules alone, but it certainly seems more prevalent in the OO way of working.

Designing any interface -- software, user, or hardware -- in isolation of the way it will actually need to be used, makes for high impedance almost every time.

And it is far easier to add to a minimal interface as (and when) new requirements become known, than it is to work around an over-specified interface that guessed wrong about what might be needed in the future. Once the latter is out there, respecify for actual new requirements, or removing the redundant, becomes impossible, because you don't know what bits someone, somewhere will have used, no matter how inappropriately.

What-ifs and maybes -- trying to predict the future and guess what applications might need -- is the ongoing curse of the software industry. Add it when you know it is needed. Never before that.