As a first approximation it is probably sufficient to do the DFT at one temperature, since the temperature in the calculation only determines the broadening of the Fermi level, to make it easier to compute the band structure. However, in principle DFT is a zero-temperature theory as long as we don't included phonons. The post-processing script then populates the excited levels according to the finite-temperature Fermi distribution. It may of course be that a finite electron temperature changes the electronic structure; I would not assume this effect to be very strong, but you want to be really sure you need to test the assumption, of course.
