Historically, the addition of OSPF on the controller was to enable HA between two controllers in different datacenters, if they were serving up remote APs. In the Campus LAN environment, it is not hard to co-locate a controller and its backup controller in the same datacenter, with access to the same user vlans; in that case, we would trunk the user vlans to both controllers, and when one fails the access points would fail over to the other, and the user would not have to re-ip. With RAP, it is understood that the controllers may be in different datacenters, so the active controller would be the default gateway for those clients, and advertise reachability for that specific subnet via OSPF. If the active controller fails, it no longer advertises reachability, the access points fail over to the controller in the other datacenter, and the new controller advertises reachability for the RAP client subnet.
In the LAN campus environment, it is fairly straightforward to just trunk the VLANs to a layer 3 switch, where the layer 3 switch is the default gateway, so that routing updates are not another thing we have to wait for to establish client connectivity. A possible routing update delay is probably more acceptable in a RAP/Remote environment, than the campus one.
I hope what I wrote makes sense....