Calculate chi-squared between model and observed spectra about csp-cmfgen HOT 3 OPEN

To be conservative, assume an error of 30% of the flux.

3% of the flux.

from csp-cmfgen.

Some notes from an offline convo with @wmwv:

1. We don't care about the statistical errors on each flux value since we expect them to be significantly smaller than the error in the calibration. To be conservative, assume an error of 30% of the flux.
2. It's not clear whether we should calculate chi-sq per band, per wavelength, per wavelength window, or per feature. Lets start with per wavelength and go from there.

from csp-cmfgen.

Pinging @spletts so she has this issue for reference.

from csp-cmfgen.