continuum normalization may fail for some cases

[megbedell]

Currently the sigma clipping in the continuum fitting function is very asymmetric, which @davidwhogg says is "playing with fire." We should check that this doesn't fail for other datasets.

[davidwhogg]

I agree.

[davidwhogg]

ps. We could clip hard and asymmetric, but then do a mean-rms adjustment to compensate for the asymmetry. Put this in as a homework question for me.

[megbedell]

maybe incorporate continuum fitting into the model itself with a GP and forget about initial normalization??