I just came across a thread that describes how the Persians added a letter to the Arabic alphabet in order to represent the French j. If ج were pronounced like the French j, they would not have needed to invent the letter ژ. I think it's rather clear that it's only quite recently that some Arabs started pronouncing ج that way. Furthermore, the books of grammar clearly state that the definite article does not assimilate into ج. If it were a fricative like the French j, it would behave like the other fricatives in the Arabic language, e.g. ش and س.
Even the affricate wasn't really common back in the 7th century (although it was already present in the Hejazi dialect, thus becoming a part of Quranic and, ultimately, standard Arabic pronunciation). Many early Persian loanwords from Arabic reflect simply [g] (as in "go"). Of course, [ʒ] must be originating from [dʒ].
As a matter of fact, in many dialects (but not in MSA) /j/ does behave as a sun consonant, assimilating the /l/ of the article. And, of course, it's not about the manner of articulation (say, both /s/ and /f/ are fricatives) but rather about the active place of articulation (coronal vs. non-coronal consonants); actually /j/ is the only coronal moon consonant of MSA, which reflects its velar past).