iCloud can resolve this problem by letting your users edit data offline and
resolving any merge conflicts automatically (and in the case of iCloud Document
syncing, giving you some options to get feedback from the user to help resolve
the merge conflicts). If this worked seamlessly, it would add great value to
your product.
If it worked seamlessly, which it doesn't. Conflict resolution is very application specific, I'm not convinced that a one-size-fits-all automated solution will ever be able to handle the edge cases as well as the original developer, who has intimate knowledge of the problem domain and the structure of the data to be synced.
App Store Features: you might get featured by Apple on the App Store for using
their latest, greatest technologies.
Don't think this is relevant for iCloud anymore. Besides, Apple has featured
tons of apps that do their own proprietary sync or use Dropbox for syncing. (e.g. Evernote, Omnifocus, 1Password). The quality of the app is much more significant than the APIs it uses.
Future Plans to move to Android / Web: a legitimate concern. But why not wait
till your iOS product has grown big enough to have substantial demand for
Android and/or Web, and develop your custom back-end solution then?
Migrating your user's data off iCloud one year later is going to be a massive, massive pain to deal with. And database syncing is not like file syncing, it won't be trivial to integrate another syncing mechanism into the application. Things will start to get quite messy if you also want to interleave iCloud in there for backwards compatibility.