But then how would you justify the "double the integral" in computing the incomplete series coefficients?

To deal with incomplete series (sin or cosine) we use complete series on $[-l,l]$.