I've been saying that all along. What happens in other languages just illustrates the possibilities. Typology, so to say. How do we know which one really take place? It would be nice at least to consider alternatives and their pros and cons. But Allen just exclaims, "look, the evidence is compatible with nasalization, surely it was nasalization!!!". What else is it compatible with? Which are the reasons to single out nasalization among the alternatives?I'm simply pointing out that how nasals behave in other languages isn't really useful for reconstructing Latin, in that there are a variety of ways this can work.
BTW, it's Tronski who points out the lack of confusion between m and n in his fundamental Historical grammar of the Latin language, even though both were "weakened" in the positions -m and -nf-, -ns-. He concludes that this may be an indication that -m retained some consonantal value.
BTW, a typological argument against Dutch-like disappearance of -m is that in Dutch as well as in Middle English, -n is more stable before vowels than before consonants.