Adam, your suggested model is actually the opposite of what I proposed. My proposal was that the distributor percentage would go down as sales went up, rewarding the person who created the product that is generating so much value. I mean, the service that the distributor provides is the same for all fonts, right? So it isn't the service that is making one font sell better than another, but rather something about the font itself. I've always held that we should maximise the financial return to the person who is creating the value, not to what become, at that scale of sales, mere rentiers.
With regard to avoiding the 'tiny royalties trap', the obvious solution would be a minimal payment to all font makers included in the collection, regardless of whether their fonts sell, in recognition of the fact that their fonts are contributing to the 'one stop shopping' value of the distributor. Call it a universal basic font income.
It may also be worth noting that the illustration of Maxim's 1963 Helvetica comes from a wonderful book on the work of Moscow lettering artists. Like a number of other examples in the book, this is a typeface that was never manufactured. Experimentation in such circumstances is cheap: it never has to undergo the rigours of making something that can be used and facing uncertain adoption in a market. There's a lot to be said for such experimentation, and one of the things I love about that book is the optimism it presents with regard to what might be possible, which ironically flourished in the Brezhnev period when so little was actually possible.
In his open letter to Vladimir Yefimov, Maxim quoted Mayakovsky: 'I love our designs’ grandeur', noting the tendency in the Soviet period to judge the success of a design by its intention rather than its achievement. We risk doing the same thing if we elevate experimentation as an end in itself.
Could I for instance use m instead of т or n instead of п without trouble?
Depends on the style of type and how consistent you are. There's no reason why an upright and formalised version of the cursive letter shape norms can't constitute a valid style (it is the norm in Bulgaria), but I would say that you should try to be consistent, so that readers familiar with the cursive shapes will engage with the style on those terms. Mixing and matching typical upright letter shapes with cursive letter shapes would be too Frankenstein, I think.
For traditional text faces, I use a simple rule: x-height + descender = cap height. Ascender height is basically stylistic, i.e. it can vary pretty freely depending on the style of type and the effect one wants to achieve in text. But the relationship of x-height and descender height to cap height is foundational. Interestingly, I didn't start from this idea, but realised it after noting that my inverted exclamation mark always fit perfectly into the x-height and ascender: I was doing it intuitively.