I don't really know. Perhaps old habits die hard. Or they want to deal with two versions. Online journals should consider remove this pointless limit and instead advise authors not to make papers longer than necessary.