The truth is, that we had some rulings on what is considered personal data and what isn't and IP adresses, even parts of them can be regarded as such. Now you and I might think this is silly etc., but if we write software for corporations that have to pay fines as a feaction of their total revenue not knowing that can easily ruin our lives.
And I am merely reflecting the way how these laws have been interpreted and ruled on so far in the comment section for an article on, well another ruling by a court. Legal reality, like physical reality doesn't go away if you close your eyes.
Is this what you originally meant?
Nontheless I hope you can see in which way the whole thing is still deeply connected to the legal question of how one can still learn about their service without tracking single users when they don't give their consent, maybe now even via DNT header.
That is personal information.
How exactly do you legally enforce "each link must not be a unique UUID generated per-user or per-page-view"? Do you mandate how "generic" a link has to be, such that a visit to that link exposes less than a certain amount of information about that visit? What about things like order-specific links on a shopping website, that will naturally identify an account if all orders share the same namespace?