Agreed. The concept of “don’t reuse your old work when you’re supposed to be creating new work” may be valid, especially in training environments, but it shouldn’t be called self-plagiarism or treated like plagiarism.

It's not just academic. If I've been paid for writing something original for a magazine or newspaper and I give them a piece that wholly or largely is just a copy of something I've written in the past (and has been published elsewhere), that's actually not kosher and they'll call you on it if they find out. Personally, I will reuse sentences and paragraphs from time to time but not entire pieces without an explanatory note.

It seems like a rule designed by journal editors to protect their turf, or PhD committees to make it easier to count original works towards degree requirements. What could possibly be the justification?

Agreed. And even etymology agrees:

Plagiarize comes from plagiarius, latin for ‘kidnapper’.

You cannot kidnap yourself

it's really a bit of a different concept in scientific publishing, not actually plagiarism. The problematic part is publishing the same results twice, because it increases the burden on reviewers and inflates your publication count. It's also just messier if the results are in multiple places since it makes it harder to follow where those results were used and cited.