The macron sign on the italic г in both Serbian and Macedonian is an oddity. The macron signs on the italic п and т make sense, because they distinguish the letters from the otherwise identical и and ш, but there is no letter in either modern orthography that could be confused with either shape of italic г. So why the macron? I am guessing that it is an artefact of cursive writing in which the single stem of the italic г could be confused with part of another letter, so is marked in this way to distinguish it from connected stems on either side. That being the case, I can see how the acute accent in ѓ would serve the same function by itself, without the macron, so I revise my suggestion above: italic ѓ doesn't need the macron.
With regard to the separation of dots from base letter shapes, this is something we did in the Aldhabi project for Microsoft, and is common in nastaliq and other cascading style fonts where the dots need to move relative to adjacent shapes. However, most software still requires that each Unicode character have a single glyph in the cmap as an initial input, so the separation of base shape and dots has to happen during the OpenType Layout GSUB operations. [Some time ago, I proposed to Microsoft a new cmap format that would map from Unicode codepoint to arbitrary sequences of glyph IDs, but they thought it would be too difficult to get a new cmap format accepted and supported in software.]
The 'four forms' analysis of Arabic script is mistaken, and really represents a particular mechanical solution to typesetting a simplified model of Arabic. In fact, what there are is two kinds of letters: those that can connect on both sides, and those that only connect on the right. The actual ways in which the letters connect, and hence how many forms they require, depend on the style of the script and what they're connecting to. There is no historical style in which there were only four forms of any given dual-joining letter. So the scheme you describe with connecting element and only two base forms may be appropriate to the style of typeface you are designing — a kind of refinement to the Simplified Arabic model of Yakout —, but not necessarily as a general model of Arabic font technology, which needs to accommodate more styles.
With regard to ligatures, I would advise to avoid them most of the time, and instead to use contextual variant forms that join in different ways, which are more efficient and more flexible.
I recall Hermann Zapf talking about the difference in the nature of his drawings for different type manufacturing processes. Drawings for mechanical production processes needed to be very precise, since they would be reproduced very directly. During the hot metal era, when machine punches or matrices would be cut on a pantograph, precise scale technical drawings were required: hence the drawing offices at Monotype and other companies that would have the job of producing these drawings. Similarly, in photomechanical processes, the typical model was e.g. for rubilith cutters to work from precise drawings and reproduce them accurately. This is quite different from Zapf's description of the earlier period of his work, having punches but by hand, in which drawings were less precise and the punchcutter was expected to interpret the drawings in the making of the punch, rather than reproducing them exactly.
I've long taken the view that manipulating bezier curves is more like cutting a punch than making a technical drawing for mechanical manufacture. If I draw, I draw fairly roughly, often at very small size, to help me figure out how to 'cut' the shape of the digital glyph.
Like much other terminology around type, I consider the term 'drawing' to be a kind of dead metaphor based on historical practice. Only in a production workflow such as Frank describes does it make sense to talk about digitising drawings in a literal sense. And yet, like Nick, I talk easily about 'drawing' in terms of manipulating bezier curves; I talk about a font that as been heavily edited as 'completely redrawn', even though no actual drawing took place; I am aware that these usages of the term are really metaphorical, just as it would be to say that the latter font is 'recut'.