Calculate gaussian noise relative to background level

[chummels]

In talking with observers, I learned that we're calculating our spectral noise slightly wrong. When adding noise to a spectrum, we currently generate a noise vector for the entire spectrum assuming a continuum value of 1, that is, by calculating the scale of the noise to be 1/SNR. Strictly speaking, the scale of this gaussian should be equal to the continuum value / SNR, so if a user has added a quasar background or Milky Way foreground, the continuum may not be 1 along the spectral range.