The development of the ending -ons seems to be pretty unclear. Grammaticalization: Studies in Latin and Romance Morphosyntax, by Jürgen Klausenburger (2002), describes it as “one of the most puzzling questions of Romance historical linguistics” (p. 51).

Klausenburger says that Corbett (1969) actually suggested that -ons may have come to be established in the present tense as a way of avoiding homophony with the perfect form derived from Latin -āvimus, which apparently is thought to have developed a geminate (/mm/, I guess, as in Italian cantammo) to “compensate” for the loss of -vi-, but then underwent degemination to produce a form with /m/ (p. 52).

Apparently, the imperfect and conditional 1plr forms used to end in -iens in Old French.