OM Components vs plain functions

I am going through this OM tutorial but it's not clear to me when to use OM components vs plain functions (in particular the om/component macro).

The tutorial writes:

The first argument is a function that takes the application state data and the backing React component, here called owner. This function must return an Om component - i.e. a model of the om/IRender interface, like om.core/component macro generates

; here the function (fn [app owner]) indeed returns an OM component
(om/root
  (fn [app owner]
    (om/component (dom/h2 nil (:text app))))
  app-state
  {:target (. js/document (getElementById "app"))})

In the next section we find the following example of a rendering loop for a list:

 ; this one does not return an om component (or does it?). it returns a virtual dom
(om/root
  (fn [app owner]
    (apply dom/ul nil
      (map (fn [text] (dom/li nil text)) (:list app))))
  app-state
  {:target (. js/document (getElementById "app0"))})

Here, we're basically just returning a (virtual) dom directly, not wrapped in an OM component, so the question would be: Why does the om/component macro exist? The macro simply helps us to reify the IRender function, but it appears that we can also just use plain functions for that. I would reify OM components that have lifecycle state (or need the owner to call get-props) but for components that just need to create virtual dom I'd rather go for simple functions (so I don't need the build/build-all functions to create my virtual dom). What am I missing here? Why is the macro still useful (and I don't see it).