If the spoke were long enough, penetrating into the head of the nipple by 2mm or more, you wouldn't be having these problems.
The nipple wouldn't break below the end of the spoke because the spoke is supporting it there, any more than a seatpost would break 2" into the frame. The general guideline for the design of nuts calls for them to have a thickness equal to the diameter of the thread (actually a bit less is OK, but this leaves a margin). Likewise a 2mm spoke should engage the head of the nipple by 2mm.
Short spoke wheels are very common because folks are so paranoid about going beyond the top, and production builders don't want to bother with too many spoke sizes in inventory. In reality most spoke/nipple combinations allow the spoke to beyond the top of the nipple a bit to allow builders to err high which ensures against the short spoke problem.
There's a certain irony to all this. For almost a century wheelbuilders used spokes that were on the long side. Mind you this was using single wall rims, so the excess had to be trimmed off. Grind marks were common on wheels through the seventies, and there were even special narrow grinding wheels sold for the job.
Bike tool companies like Var and Eldi sold
nippers made expressly for the job of cutting excess spoke flush at the head. Later on with the advent of hollow rims and excess spoke beyond the top of the nipple no longer mattering folks started going the other way and erring low. Part of the reason is that many spoke/nipples have less room to err high before the thread jams, but that's solvable by good selection of materials.
In any case I'm of the opinion that any wheel where nipples break where spokes end below the head of the nipple is inherently defective, though good luck getting manufacturers to accept responsibility.