Looking at this from the point of view of an editor who handles a paper, the major important issue here is the quality of the resulting paper. If reviewers' comments make sense, it is irrelevant what these reviewers do in their own papers. That's just a different editor's job. If a reviewer asks for something that is important for raising the quality of the paper, this is a valuable request, and if the reviewer doesn't do that thing themselves when they publish, another editor and their reviewers should ask for it. They may not, but I have no influence on that.

On the other hand it is true that sometimes reviewers ask for things that may in principle have some value, but may require a huge effort for little effect, or would be better addressed in a different paper as the present paper is already long enough and only so much can be expected from a single publication. Once more, from my point of view, it isn't really relevant in such cases whether the reviewer would do such things in their own papers. Rather I'd expect the author to say in their reply to reviewers why a certain change/addition is not made. As an author I rather regularly do such a thing myself. Sometimes the editor is just fine with it (remember, it is not the reviewers who decide whether a paper is accepted, it's the editor). Sometimes the paper goes to the same reviewers again and it may happen that a reviewer is not happy that one of their issues has not been addressed. In this case they may insist and give arguments why they believe it's essential, and the editor may or may not agree. There may also be a path to compromise in that the reviewer still insists on something that was requested before but not all of it, and then the thing may go into another round.

The main point however is again that it doesn't matter for this what the reviewer does in their own publications. If you as author believe that certain requested changes are unreasonable, be it because you don't agree, be it because the effort would be unreasonable or it would go beyond the scope of the paper, you can defend your point of view in the reply to reviewers, and then it depends on whether and to what extent you can convince the editor and the reviewers - and it can happen for sure that an editor will accept something even if one of the reviewers (who may have asked for something unreasonable, and maybe for something they wouldn't do in their own papers, but that doesn't really make a difference) remains unhappy.
 
Regarding the more "empirical" side of the question, I've got to say that as editor I'm not usually interested, i.e., I will not check reviews against what the reviewers do in their own publications, because I think it's irrelevant to my job. However of course I often nominate reviewers whose work I know to some extent, and I rarely have seen strong discrepancies between reviewer demands and standard of their own work of the kind the question asks for (as a statistician, I do occasionally see reviewers asking for more simulations in methodological papers who try to get away with a minimum in their own, so it does exist to some extent).