For those Elsevier authors who wish to provide OA rather than continuing to agonize over what Elsevier might intend or mean:
Believe Elsevier when they state officially that "Elsevier believes that individual authors should be able to distribute their AAMs for their personal voluntary needs and interests, e.g. posting to their websites or their institution’s repository, e-mailing to colleagues."
Go ahead and deposit your final draft immediately upon acceptance for publication, set access to the deposit as OA, and ignore all the accompanying Elsevier hedging completely. It means absolutely nothing.
And for those who nevertheless remain tormented by irrational doubts:
Don't stress: Deposit immediately just the same, but set access to the deposit as restricted access (only you can access it) instead of OA, and rely on the repository's copy-request Button to forward individual eprint requests to you from individual requestors: you can decide for each request, on a case by case basis, whether or not you wish to fulfill that request, with one click.
This will tide over potential user needs till either the Elsevier embargo elapses or your irrational doubts subside -- whichever comes first.
(The battle-ground for OA has now become the 1-year embargo, which publishers try to impose in order to protect their current revenue streams come what may. Publishers -- though so far not Elsevier -- have tried to redefine Green OA as access after a 1-year embargo, leaving authors who want to provide immediate access with only one option: pay extra for Gold OA. The immediate-deposit mandate plus the eprint-request Button -- not petitions, boycott threats or hand-wringing -- are the way the research community can protect the interests of research from the self-interest of publishers.)