the original word DIDN'T have a spoken 'r' in the second syllable when it came into the language, so it is extremely unlikely for the 'r' sound to have been added, since, as you point out here
I haven't looked at the citations, but how do we know this? All sorts of things happen to words after they've been borrowed. Middle English parfit became Present-Day English perfect. Somehow, I doubt it has anything to do with rhotic and non-rhotic dialects or orthography.
True, strange things happen in English. But, since both parfit and perfect have an 'r' in them I don't quite get your point there? But then it is late here and I'm probably missing the obvious.
I still think it's far less likely for an 'r'
sound (as opposed to merely the letter) between a vowel and a final consonant to be added than to be subtracted. Assuming the original assertion that it's not there in the Arabic word is correct - I never checked that.
