No, it's a rule that you find very often in Old, iambic poetry which may even shorten syllables that are long by position. I'm too lazy to look up examples right now, but I can find you some if you insist
I don't think you will find too many examples of -erunt having a short e, at least I didn't come across any when I looked up the passages above. In most cases the natural stress would be on the e, so you cannot apply that rule. It works in Cato's example though, since the addition of the -que puts the stress on the u.
Anyway, I find your attempts to reconstruct classical pronunciation from modern Romance languages quite interesting, but I'm not sure too which extent it works. Even if you could prove or provide strong evidence that the e was short once, it could still be possible that it was long at Cicero's and Caesar's time and shifted only later (or shifted differently in different dialects ... vowels are quite instable after all).
btw my Latin professor has quite strong feelings about the -erunt. People saying
laudáverunt is one thing that really drives him mad (along with
Caesáris and
non with a short o)