That would mean sacrificing basic scientific principles - "cite at least the most relevant stuff" - for political reasons. I don't think this is a good way for improving scientific publishing.

In some cases that can be true, but with the current shotgun approach to academic publishing, the authors have often republished variations on a paper, covering slightly different aspects with different framing, in 3 or 4 different venues, so sometimes there'd be scope to choose to cite a non-Elsevier one.

Ok. This means that even more authors might republish the same research in different venues - just in order to avoid their research being ignored because of their choice of journal. Seems like another danger of an Elsevier boycot to me.

That's not really something I would want to try. Omitting references can lead to accusations that you are trying to claim as your own novel contributions things that others have already done.

You could reference another paper that discusses it (assuming there is one), perhaps one that references the paper in question, ideally by the original authors. It's common for authors to write a series of papers on some topic.

This doesn't give the ideal credit, but it addresses the concern you raise.