Which review recommendation should I give to maximize the likelihood that a manuscript is rejected?

Is this sound and appropriate?

No. Your job as a reviewer is not to choose which manuscripts are accepted or rejected. Your job is to advise the editor as to the merits and the flaws of the manuscript, and then let the editor decide the fate of the manuscript.

Thus, you should give the manuscript the ratings that most accurately reflect your perception of it, not the ratings that you think will get it "killed".

I think that there is an underlying misunderstanding

I don't think I misunderstood the first version of your question, and my answer still applies. It is not up to the reviewer to maximize the likelihood of any particular outcome; just tell the editor what you think of the manuscript, and leave the rest to the editor.

This means that if you are asked by the editor to recommend an overall decision, you should recommend an overall decision. If asked by the editor to recommend a decision based on quality and specifically excluding the priority, novelty, and usefulness of contribution, then you should do exactly that. (If you aren't clear on what factors should be included your quality recommendation, ask the editor.) If you believe that a manuscript should be rejected from a given journal because its contribution is incremental, you can indicate as much in the part of the review where you're asked to assess the manuscript's originality and/or in a separate comment to the editor.

If your intention is to "kill" the manuscript, you should give a reject recommendation. Rejecting does not mean there are no merits to the manuscript. In fact, I would guess that most manuscripts are rejected because the journal is not a good fit, or the results are not strong enough for the journal, but might be appropriate for a lower tier or more specialised journal.

Of course you can and should discuss the merits of the manuscript in your report, as well as suggestions you think would improve the paper, even if your recommendation is to reject. It is then up to the editor to make the final decision.

In my opinion, you should only give a "revisions" recommendation if you will likely accept the manuscript after the requested revisions are made. Purposefully setting the bar for these revisions too high because your original intent is to "kill" the manuscript for that journal seems disingenuous to me.

There is no "kill the manuscript" in a reviewer's job description.

If the reviewer thinks the manuscript is unoriginal or flawed (e.g. technically), that's what they should say in their review. They can recommend it to be rejected for the given journal.

However, the way the question is put sounds as if the reviewer may want to try to manipulate the editor, and if they do, they arrogate to themselves a role which they do not have.

Their job is to identify whether the paper is sound, original, (and possibly) fitting to the journal, according to their best knowledge and in good faith. A scientist thinking in terms of "killing manuscripts" effectively excludes themselves from the pool of available reviewers, hands down.