Fair questions. I've addressed a number of these in existing articles concerning information and markets. A search of my subreddit turns up most of these:
https://www.reddit.com/r/dredmorbius/search?q=information+ma...
Specific to your questions:
1. The problem with per-article paid access is that the question of access is tied to the question of cost. I'm working at a scale of information where I'm looking at thousands of books and articles. The overhead is nontrivial.
A library model -- where you simply have stacks access and don't pay per item viewed -- works out quite well. A rental model, where you're aware that each item carries some access cost, however low: $10, $1, $0.10, $0.1 -- adds up. For much of the world, any per-item cost is high.
The Internet offers us several such "library" models --Sci-Hub, LibGen, and BookZZ each provided unencumbered access to massive troves of published material. Extralegally.
Yes, you might find some way of wedging a paid-access metering model onto this, but to what end? Consider that the average American spends something on the order of $200 on direct media purchases (books, magazines, music, video) in a year, and an indirect $600 or so in advertising. Which has all kinds of distortionary effects on what is available, or produced. There are ~300k writers and editors in the country, whose core need is to a) keep a roof over their head and food on the table and b) have the tools for writing available to them. Seems to me that cutting out the inefficiencies and distortions of advertising would be a net win.
Your 2) is false: at some point, I've got to consciously decide whether or not I'm paying for content. Otherwise you're making a sham of the concept of a discretionary payment system.
For 3 & 4: it depends on what the granularity of the system is, on both payments and units of consumption.
First, with Patreon, you aren't paying for access to individual items of content, for the most part. The models vary, but range from "I'm pitching in to support someon I believe in" to "I'm gaining time-based channel access".
There's a word for that last, it's call a subscription. And subscription-based media access ... is very much a thing.
There's an entire industry that grew up around this, at a previous time when per-item access to media was expensive, and the joint problems of "how do I sample the wares" (for audiences) and "how do we keep our authors fed , housed, and clothed" (for broadcasters) came up.
The answer was the literary or publisher's magazine.
This was a regularly-produced (weekly, monthly, quarterly) publication, with a stable of writers, a compilation of works included with each issue, often publishing longer works in serial form, etc., etc.
It served as an advertisement for long-form books, should subscribers choose to buy them. It mitigated the risks (and bundled purchasing decisions), at the level of multiple authors (anywhere from a handful to 100s), on an annual basis, and generally offered something in any given issue worth reading.
It wasn't until the 1860s or so that additional advertising (beyond the books and publications themselves) became a significant component, though that grew rapidly through the last half of the 19th century. Hamilton Holt's Commercialism and Journalism, 1909, covers this in a brief 115 page booklet (available at the Internet Archive). (Holt was a magazine publisher, reporting on the industry.)
I see the present situation of one in which the question of where and how the bundling size has changed, and is shaking out. The previous regime served the needs and limits of paper-based publication, as well as vinyl & CD-based audio, and VCR and DVD-based video. In a world of ubiquitous digital distribution, those limits and gateways no longer exist.
And information is a public good, as Google's Chief Economist argues.