Since entity views can include rich text fields, is it primarily the ability to easily interleave text blocks and view blocks that makes a document preferable?
Yes and something else: if I have a new team and I want to “document” the entity view with some rich text fields, either I have to enter the texts manually or we use automation to fill them automatically (after all they are the same for each team) but rich text automations are really a pain to use (Copy-pasting and markdown interpretation in Rich Text automation actions). A document can be duplicated. It’s not ideal either since the embedded views have to be changed but it feels less clunky (maintaining automations add technical mental load on the long term).
For me, an ideal solution would be a “templated document” that you could add in a Smart Folder where you could reference in text blocks the fields of the entity setting the context and in which you could embed context-filtered views.