I think the idea is that one would have a "source" -- say a book -- and that subsequent citations of it would only involve typing in a page number.That is, it was supposed to be a time and effort saver. I fear that the source/citation bifurcation has mostly served to confuse thousands of people.
I understand and agree with your analysis of the genesis of the conventional source/citation distinction in genealogy. I just wonder if the distinction makes sense anymore, or if it ever did. Also, the "page number as the citation" concept was early on generalized to census records, birth records, death records, marriage records, and really to all other records. But many times such a generalization doesn't make any sense to me. For example, if you get a pre-prepared footnote sentence from a genealogy Web site (and most of the main sites do offer a pre-prepared footnote sentence you can copy and paste), about the only reasonable way to use it in RM is to use a free form source template and paste the pre-prepared footnote sentence into the "Footnote sentence" field. This makes the entire RM Master Source as source and Source Detail as citation concept pretty meaningless.
As a still fairly new convert to complete source splitting, I don't find the absence of the ability to type in just a page number to be any kind of problem at all. Instead, I just go into Lists>Source List and copy an existing Master Source that's nearly identical to the one I'm about to make as a way to begin the new one. Now, that's a real time saver. What I am missing, however, is the ability to produce a bibliography. Bibliographies work great for books. But I wonder if the conventional notion of a bibliography really works very well with most genealogical sources. As a thought experiment, suppose I wasn't using RM or any other genealogy software and suppose I was using Microsoft Word to produce a family history book. And suppose my book was full of footnotes that were citations to my evidence. Under these circumstances of doing things totally manually and having total freedom to do things "right", what would a proper bibliography look like that referenced census and birth and death and marriage records? I don't really know. I do know that under these circumstances there really isn't any concept at all of source lumping or source splitting because I am simply hand coding all the footnote sentences.
Finally, I know you are a GedSite user, as am I. GedSite embraces a pretty standard genealogical view of the distinction between what a citation is and what a source is and it supports both citation lists and source lists. In my case, GedSite is getting its citation and source data from GEDCOM produced by RM. The GedSite documentation specifically says that a GedSite source list is not the same thing as a bibliography, although it sort of seems like a bibliography to me. In any case, I can't figure out what the function of a source list in GedSite would really be and how it would differ from a citation list. In my case as a source splitter, the citation list and the source list seem identical. But seems to me that the lists would be identical even if I were not a source splitter.