Unless there's a specification in a design somewhere (not in my world....), I put the diameter to nominal, minus 10 percent of the thread pitch. Then I'll go down the thread with a triangular file, leaning on the "corners" of the flat points of the thread, kinda making a "faceted round crest". They come out well within spec for most anything outside of a nuclear reactor.
I will use the file to "ease" the corners of the flattened crest BEFORE taking the last couple of final cuts, while I'm matching up a nut, since in my world it's probably a commodity grade nut, which means it might or might not be inside or outside of any specification anyway, at the manufacturer's leasure..... That just makes sure that I don't interfere with their rounded root. (rounded and plated, probably dirty before plating, and generally inconsistent root....). Then I get the fit that I want, since it's probably "custom" anyhow, since I don't make threads for sport, when there's a dirt cheap and very suitable commodity item available for less than the cost of materials. And plated too...
Depth of cut? It depends what you're driving. On my lathe, your depths sound plausible. After a "scratch pass" to make sure I didn't bollix up the gearbox again (which let's be clear.... It's very intuitive, and clearly labeled with a metal tag in good condition. No excuses here), after the scratch pass I might start at 12 or 15 on the first pass (Depth of cut, radius, double that for diameter reduction), but they drop quick after that. By pass number two, it's "dug in" more, five sounds about right. Nearing the depth, two will deflect less, and give me a straighter thing for the tweaking in passes. The last pass (or first of however many "test fit" passes" I might aim for less than two. One maybe, or even splitting a thousandth to some fraction depending on how close I guess that I might be. If you're not bound up on the roots or crowns, then the smallest whisker really goes a long ways towards making them fit "just right".
