There seems a mismatch between the "in context" ambition and what seems like a "Series of lessons that experienced Wikipedians would demand very new WIkipedian know before they are let loose". Nobody is going to do Modules 1-500 if they just want to fix a spelling. I'd be more inclined to ask each new user (based on some minimum number of edits) first "What do you want to do", offering the likely options like "fix spelling, grammar, punctuation", "change something that is wrong", "add some new information", "start a new article about me, my business, my product", "start a new article not about me, my business, or my product", as you want to teach them different things. Obviously there are more things they might want to do. Then tell them the *minimum* they need to know to do that task (just-in-time learning).
The change/add need to know about citations for their new information. I wouldn't give them the full range of citation-providing formats. E.g. in the VE, I'd just give them a text box where they can paste URLs or type whatever and emit it wrapped in ref tags and let someone else sort it out.
Those wanting to write articles about themselves should be sent packing, I doubt that this type of person is likely to become a regular contributor (too self-motivated), so let's get rid of them quickly before they waste everyone's time at AfC.
Those wanting to write articles not about themselves etc (no CoI) should firmly be told "not now, it is too soon, you need at least 100 edits to existing articles to gain experience" (this is what I say in F2F training and I am very upfront that the unlikelihood of a new users successfully creating a new article - I am an occasional reviewer at AfC and I know what it's like). AfC involves the user making a lot of effort and seeing it never published; not a useful way to onboard someone so just STOP them from doing it. AfC is a stupid solution to the problem. Once they've met the magic 100 (or however many) edits, when you ask what they want to do, offer "start a new article" for the first time and talk about notability and how to demonstrate it. Move to a *lightweight* AfC process where initially they fill in a form answering "title", "what is about", "what is it important/interesting for Wikipedia", and 2 independent citations (stop them investing a lot of effort on article writing until they have the green light on notability) - including some examples of "good" and "bad" ways to answer these questions in terms of demonstrating notability.
While we have watchlists for articles being edited, we don't have watchlists for "new user doing edit". I have a topic interest (Queensland). If any new user turns up editing on that content, I am happy to reach out and offer to help (having some canned ways to do this helps a lot - I use Twinkle > welcome > project-specific WikiProject Australia as my standard welcome). If there was a way I could sign up to be notified of the activity of new users within my topic space, that would be a good thing (for this purpose, topic space could be defined either by a category closure or a WikiProject template). At the moment, they have to edit something on my watchlist for me to notice them.
My own experience of supporting both F2F and remote new users is that email is infinitely preferred to Talk or Teahouse. Firstly they understand how to send an email, they don't understand how to write on Talk or Teahouse (assumes you know about source editing! - why no VE on Talk, User Talk or Teahouse). I put a VEFriendly template on my User Talk page so they can use it if they want to do so with VE, but really they like email. Also email is private (their ignorance is not public knowledge and they can be honest about what happened if it involves another user) and they can email screenshots, which you cannot easily do on-wiki). While we know that there is a an "email this user" on left side-bar of a user page, it's rarely found by a new user (and of course some users don't have a registered email address on their Wikipedia account). I hand out business cards at F2F training with my email address and I publish my email address via whatever medium is being used by remote training. Because I edit with my real name, I actually get a lot of new users where I have written something on their User Talk page (or sometimes a potential new user who has found my name in the history of an article or an article talk page) tracks me down in real life (fairly easy to do as I have a website on which I have both my real name and my email address) and then emails me or phones me or manages to contact me via a friend of a friend from Linked-In. It shows how hard our Talk mechanism is that new people find it easier to reach me by such other means! I think when we ask the new user "what do you want to do today", I think contacting someone they have encountered or someone connected to an article of interest to them should be an option (might require a bit of detective work on their user page, or recent edits on any articles they are interested in).
As part and parcel of this, we must very strongly encourage the provision of an email address when signing up, and keep suggesting it while they are a new user, as it is the best way to talk with them.
The other thing I think we should do is divert any new user from making a change to a high quality article or high readership or one with many page watchers. These are very high risk places for new users to start. Instead just put their request onto the Talk page and hope someone follows up with them. Only allow them to do early edits on lower-risk articles, low importance, stub/start/C quality, low readership, low page watchers. This is the approach I take in F2F training, I start them working on their own User page to learn the minimal skills (bold/italic/heading/link/cite) and then provide "low-risk" real articles and one or more sources for each of them to update the article. I was a mentor in the OCLC course for around 300 public librarians last year and I saw that many of the new users went for "high risk" topics for their first edit and were reverted. I don't know if anyone has the revert stats based on the risk factors I mention, but I would be very surprised if revert rates aren't higher on the articles I identify as high-risk. I know when I edit outside my normal topic area, I experience more reverts, not because I don't know standard Wikipedia rules but because I don't know what that topic area regards as a reliable source, or what conventions/agreements have been established at a particular WikiProject, or even how that sub-community interpret the same standard rules.