error and data quality handling in dark current subtraction

[zonca]

@ikashell @arunsurya77 currently, the dark current subtraction step
merges the data quality flags (with OR) coming from the dark current file stored in CRDS,
I think this is reasonable, please let me know if not.

Instead, if there is a error extension in the dark, that is unused, should instead combine the errors in quadrature? or some other way?

[arunsurya77]

I would think the error propagation is just adding by quadrature for each pixel in this case since its just a simple subtraction. Will check on this