One simple criterion is that if you have refereed more papers than the number of referee reports you have received for your papers, perhaps even divided by the number of coauthors you have for each paper, then you are ahead of the game and shouldn't feel guilty about turning requests down. Ultimately, if your career goes really well you should expect to do substantially more than your share of reviewing, but you aren't obligated to do this while still a postdoc.
Even if you haven't reached this point, it's reasonable to put your career development first for now. Strategically, you should referee a paper if reading it will be valuable for your research, if you can impress an influential editor, or if it's for a particularly prestigious journal (so it will stand out on your CV in the list of journals you have reviewed for). Beyond that, you can do what you reasonably have time for, but you should feel free to decline review requests when you are busy with other things. If you feel guilty about this, you can make up for it by reviewing more papers than you would otherwise like someday when you have a stable job.
At what point will listing another journal (or another instance of refereeing for a given journal) on your CV generate negligible returns for an early-career researcher?
I'd say the returns become negligible pretty quickly. Adding a really fancy journal can look impressive, but even doubling the number of run of the mill journals will make only a small difference. (People want to see that your reputation leads to review requests, but don't care much beyond that.)
To the extent refereeing helps your career, it's more likely to be because you impressed the editor than because a hiring committee liked the line on your CV. However, writing papers and giving talks will reach a much broader audience than refereeing can, so extra refereeing is not an efficient career boost.