My surprise at not finding it was mainly from Scriv being SO flexible and ingeniously kitted out that I couldn’t believe it doesn’t do this particular trick.
It really has less to do with strategy and more to do with technical limitations, although I do hope that some day there will be some opportunity to circumvent that.
That said I am and always will be a staunch proponent of text-based solutions to problems of this nature (as I am with general formatting as well). Text-based links never degrade. They don’t get broken when you copy and paste into a different program. They won’t be lost when the software you used to create them goes out of business. They will work as well today as they will in 150 years, when your estate is publishing your working manuscripts to quantum ebooks or whatever (although with the way the climate is going, stone tablets may be more likely 
).
This will probably recondition my thinking more toward the high granularity Scriv likes, and away from many decades of the word-processor semi-monolithic mindset. The hierarchy will still be there, but I won’t be oriented toward the Binder as the main mechanism of entry.
I’m not familiar with Lotus Manuscript, though it sounds very interesting and worth reading up on (as many of those older Lotus programs were, in their own ways, groundbreaking).
The hierarchy was used mainly to control formatting and content, but one seldom thought about the hierarchy except when defining formats for each level.
That is a good description of how Scrivener is intended to be used, in fact! With its Section Types feature you can map hierarchy to formatting, where structure can become expressive in that same sense as Manuscripts did.
This is not uncommon in outliner based software, where I find Scrivener more interesting is in how it can subvert that otherwise rigid approach. Most outliners have strict formatting rules based upon hierarchy and no hidden text, but in Scrivener we can break those rules in multiple ways. We can even have hierarchy that intentionally has no expression, that looks to the reader like a single page of text, or as notes, doesn’t print at all. It allows us to think about the text in a fashion that may go deeper than how we intend to present that text.
I’d suggest downloading a copy of the user manual project to see an example of how structured a large document can be in the binder. You will encounter sections that work much like how you describe blocks, I believe, such as the menus appendix, where each menu command in the software is its own binder item. These have roles similar to how you describe Manuscripts, where a block has a meaning, a way in which something is formatted ultimately.
And yes, as you have concluded, Scrivenings mode is essential to text that is so highly structured as that. I leave the Binder largely collapsed, only opening things up to the chapter level—a +3k item outline would never be practical in so compact a space. One way in which I work with a chapter’s content is to click on it in the binder, where it loads into an Outliner view on the left, fully expanded (with sometimes upwards of several dozen outline elements), and simultaneously a Scrivenings view on the right (you can achieve this with the Navigate ▸ Binder Select Affects ▸ Both Editors setting). I furthermore link the left outliner to the right view, using the paired navigation setting below that one, so that I can narrow down the text to the section I’m interested in, causing it to act a bit like a second binder. This is useful in those longer chapters, like the one on the compile format designer.
Now what you won’t see much of in that project are point to point links. I do use them, even though most of the cross-reference style links take me right to where I need to be without scrolling. I do use point linking for editing notes, which all get stripped out of the public copy. If I need to highlight a specific sentence, or even word, for a particular bug or feature report, then I’ll use inline annotations and markers for that. One bug has one unique marker, which then may appear in a dozen different places. Searching for that marker reveals all of the places that need revision when the bug is eventually fixed.
Perhaps some programs have ways of working with hyperlinks in a similar fashion, but it is very unusual to see much more beyond rudimentary click and scroll behaviour.
Meanwhile, one quick (final) stab at defining the functionality I still hope may turn up one day – an emulation (if you will) of the Comment facility in Adobe Acrobat (or Reader). One hilites some text, and a blank comment (no title needed) appears in the right pane. Later, with the read cursor anywhere in the file, one clicks any hilite and the Comment pane scrolls to the appropriate Comment. Or, one pokes around among the comments, clicks on one, and the read cursor jumps to that hilite to center window.
Well that you can already do just fine! In fact that was one of the alternate techniques I described in the other post. I prefer inline annotations because I like using the Quick Search field in the toolbar to jump straight to the marker, and that works better if the marker is in the text. But you can also put the marker into a Comment in the sidebar, where they work just like you’re talking about, you just need to use the mouse more. It’s also worth noting that the Comments tab in the inspector will collate across the entire Scrivenings session.
Oh, and I forgot to mention, but as for getting back to where you linked from, that is of course what the Back button is for, or the Ctrl+[ / ⌘[ keyboard shortcut. That will take you right back to the cursor position you left off with, which if you’re using this marker-based technique, will mean with the marker selected.
Lastly, thank you very much for the kind words! There is still a lot of work to be done on with Scrivener, but it sure is nice to finally get the fuller vision for what this software was meant to be out into public use.