I agree with the many other who think #1 is the way to go. Or rather, a variant that I'd call #0: there shouldn't be any pretending you didn't notice. Acceptance of a paper should IMHO solely depend on scientific merits and content (includes whether the subject fits scope of the journal) plus possibly pre-specified fees. Pre-specified fees could also be a payment in kind, i.e. saying that for every manuscript you submit, you have to do so many reviews.
As for reasons not to review, I don't see how "I cannot afford to put in the time" is worse than saying "I don't have the time" or just not giving any reason. If you consider "I cannot put in the time without being paid" not a good reason, then surely neither is no reason a good reason.
In my field, academics are usually employed. And the employment contracts I had in academia always considered not only writing up research into manuscript form but also reviewing as part of the professional duties. In that sense, I've been paid for the vast majority of reviews I've done. Just not by the journal. BTW, there are macro-economic estimates of these costs (STM report 2012, p. 21 last paragraph
If I were journal editor receiving such a request, I'd be tempted to say "Please send us your/your employer's billing address and VAT number otherwise we cannot prepare a reverse invoice. If you act self-employed, we'll need your VAT number or VAT exemption and tax number. We'll then be considering what we can offer."
And I do see the point that a publication system that pays reviewers, collects fees for submission and reading and pays out royalties to the authors (or their employer, in case of employment contracts) may result in almost a zero-sum game in terms of money (even if we'd divide all the profits of Elsevier, Springer and Wiley) - at the additional cost of huge burocracy. But it would lead to an immense increase in common academic knowledge about international tax rules... Even though you can count me as one of the very few people who are not immediately scared off by the fact that receiving fees does cause tax payment and reporting duties, I'm not sure the gain in fairness of even an ideally fair system of those fees is worth the hassle.
I'm freelancer now, but still have some projects that are very close to academia, and still do some research. Let me add a point of view from that perspective, because that's where I can see me being the reviewer in your question. I'd have explained, though, that as opposed to people being paid by grant money I'm freelancer and do not have any project covering the review of your manuscript - and that at present I cannot afford volunteering the time for the review. Taking the scenario further, as an author getting response #2 I'd answer that I feel at least as magnanimous as I'm not paid by any project for preparing the manuscript*.
I'd like to point out that I'm fine with the current system if there is a project that at least somewhat covers these activities (I do have one such research project right now). And I'm fine with volunteering time even if there is no such project, but within limits.
But I do have to say that I have some disquieting experience with academics on permanent employment contracts who do not see any more the difference between having and not having an employment contract. (If I'm the one in question, that's fine - I can deal with that professionally. But I do get upset if this attitude hits former students who are out of job and are expected to work for free.) And that alarm is triggered by seeing many responses here on academia.sx that claim pretty much noone is paid for reviewing when my estimate is that the vast majority of academia.sx users do have academic employment contracts which include publication duties (please comment that you are not paid by projects if that's the case so I can update my world-view).
* I'd rationalize to myself some of that time as going into marketing. And in fact, I would not try to haggle for payment, but do foresee me haggling with journals whether if I volunteer my time for writing the manuscript, they could throw in open access. Or leave me more rights than usual.