Dave, one of us is confused, and I’m not sure which of us it is. You say:
Luncheon is attested earlier than lunch.
But in the linked article, you write (quite correctly, in my view):
Even though luncheon appears later [read “earlier"] in the record, an eleven-year difference across a span of more than four centuries is not that great, and the shorter lunch could be the original form. The bottom line is we don’t know which came first, but the best guess is that it was lunch.
Is the gap significant or isn’t it? And if lunch was (per “the best guess") the original form, how is luncheon relevant?
Then you say:
The connection with lonja doesn’t come along until later, and only happens to be made because lunch is used to translate lonja in a bilingual dictionary—which also kind of kills the derivation from lonja because why would the dictionary editor have used an unfamiliar borrowing to translate the root word? One would expect “hunk” or “thick slice” if lunch were not already familiar to the editor and readers of the dictionary.
With respect, that doesn’t make any sense. Lunch could a) be already familiar to the editor and readers of the dictionary, b) be borrowed from lonja, and c) be used to translate lonja, all at the same time. I don’t know why this is so difficult; cheese is borrowed from Latin caseus, and yet if you look up caseus in a Latin-English dictionary, it says… cheese!