Ironically, I’m not the only one who’s done this.
This thread post-dates my own efforts by over 26 months, but, most pleasingly, echos my own logic to a most satisfying level. He states he already has a high speed lathe function, and has chosen a differing harmonic drive, but his approach has all the checkpoints mine did, and more (such as testing the harmonic before use). Even better, he is a PhD in measurement, and this reflects well on my thought processes.
There is excellent information on bearings, building philosophies, engineering trade offs, final operations.