Resolve minor discrepancy in Calzetti 2000 attenuation

[aphearin]

The DSPS dust models are implemented in the attenuation_kernels.py module. The unit-testing for our implementation comes from values I computed using the public dust_attenuation package, and DSPS agrees with that implementation of Calzetti 2000 with machine precision. However, when I compute the impact of dust on the SSP spectra using FSPS in this notebook, I get a small disagreement. We should dig into this a bit to see if we can get to the bottom of the small discrepancy.