I'll give this a look over the weekend!
I believe I'm in the middle of changing some code related to mask modes but something like *this* is really odd... It might have to do with how I'm caching materials...? I may not be checking if the mask value is the same with material caches (STM will have two objects share a material if the base material, font, and filter mode are the same), so that could cause this to trip up! Does the error still happen if the STM objects in hierarchy branch A are given a different font? This would basically confirm that this is the issue for me!