I've looked for ways to make mathGlyphs easier to translate. Adding support for transformation matrices would be nice but it would be quite complex.
What about this notation to move all the geometry in a glyph by (100, 200)
a = glyph.toMathGlyph()
b = a + (100, 200)
This would be analogous to the support __mul__ has for anisotropic factors. __add__ and __sub__ can check if the object is a glyph or a tuple and then call _processMathOne or a new _processTranslate.