Are the Hardy-Littlewood “rules” unethical?

I've always interpreted such remarks as meaning that the individual papers were just progress reports on their larger enterprise, to which both made irreplaceable contributions, and that they had no interest in keeping track of moment-by-moment relative contributions.

In particular, "on average", there was no misattribution or false credit.

(Further, the "not contributed the least" may easily be hyperbole, just to make the point.)

For that matter, if one is in regular correspondence with another, how to attribute ideas that develop gradually? I think their solution was entirely reasonable.

I think the context of the rule is important here. The rules taken together set up a system of trust: that the authors trust each other to contribute fully to the project, and so (as others have pointed out) they didn't have to waste time with the nitty-gritty of specific contributions.

It is not useful therefore to view this rule in isolation as a license to willy-nilly add authors without contributions.

Generally speaking, this practice would not be acceptable under today's ethical standards.

I don't think these rules represented standard practice even in Hardy and Littlewood's day. They wrote them only to govern their own collaboration; I'm not aware that they ever even suggested that anybody else follow them. The rules are notable because they are unusual (and, as mentioned by Anonymous Mathematician, humorously exaggerated).

If your reputation matches that of Hardy and Littlewood, you may find the academic community (and your institution) willing to tolerate idiosyncracies like this. Otherwise, I wouldn't suggest trying it.