Switching to SSD2

[twshimwell]

Whilst imaging some galactic fields I was comparing SSD to SSD2 as thought we might as well migrate to SSD2.

For the field im working on the models produced by SSD and SSD2 look very similar when cleaning with comparable parameters and the same mask. Both leaving comparable residuals (first attached image).

To my surprise though if I do the DI calibration step on these images then the resulting dirty DI calibrated image using the SSD model looks much worse than that from the SSD2 model (second attached image where right is when the DI calibration uses the SSD2 model and the left is when it used the SSD model).

It would be nice to figure out why it is so different... @cyriltasse is there any difference between the SSD and SSD2 predict?

Also would be nice to switch to SSD2.