The issue is that that Ohanzee has a different goal from Kohana, and from what I think many of the developers want. I (and I think many people here) want a framework that has all the major pieces that work together well with the ability to pull in other components / libraries easily. It's not important to me that the Kohana ORM can be used in another project but it is important that I can easily use Doctrine or Idiorm when that fits my need better than the Kohana solution.
I can appreciate why someone would want to work on a library that was loosely or non-coupled - it increases the chances that it'll get used.. But it's unimportant in my decision on choosing a framework.
If anything I view it as being a drawback. If I have one framework that solves most/all the problems that are typically needed in a web application, then I can now look for other developers with knowledge of that framework, or they could look at the documentation or go to SO to ask about it. With decoupled components I have multiple coding/config styles and documentation in multiple projects to follow.
As I mentioned, I think it is valuable (and one of the strengths of Kohana) to be able to easily import modules/components when needed, but at least then it done for a specific reason because I need something specific that Kohana doesn't offer (e.g., I want to be able to use mongo or mysql - so I choose doctrine because it supports both, Kohana ORM doesn't) If you're replacing most of the components then that person/team would be better off with a component based approach like Ohanzee or Aura. But I think the average developer doesn't need that and the consistency of having all the pieces without having to bundle them themselves is desirable.