Chad, who made the original post, is a colleague of mine. I made a proposal to the editor in question, trying to implement this system informally by asking the Section Editors just to send an e-mail with their decision/recommendation to the Editor but NOT to record the decision using OJS. It hasn't worked for this journal. The Section Editors find it a very natural thing to record the decision using OJS.
The crux of the issue is the key words in Chad's final bullet point "and informs the Author." Right now as soon as the Section Editor records the decision using the system, OJS automatically sends to the Author the EDITOR_REVIEW prepared e-mail communicating that decision. Unless we don't have OJS configured properly, I see no way of disabling that e-mail. And, even if the e-mail itself could be disabled, the decision is recorded and the Author can see it any time he or she logs into the system to check the status of the submission. Having the Editor later reverse the Section Editor's decision (which, admittedly, the system very easily allows) puts everyone (Editor, Section Editor, and Author) in an awkward position.
Did you have something different in mind than what I've described when you wrote that "the workflow is easily possible informally" that would avoid this potentially embarrassing scenario? Are you aware of any users who may have made the type of modifications Chad describes.