I'll defer to someone who has actually done the comparison and worked out which way is best, but I reckon you
could do either... but...
Hmmm... thinking about it a bit more...
If you plan to stack them but
average them in DSS then just stacking them in DSS would be fine. However, usually DSO images are
added to bring out the faint bits... so if you add your 200ASA and 800ASA images, the burned out highlights in the 800ASA images will burn out the final image.
So I would suggest, stacking your images by exposure in DSS then HDR processing in PS. But hopefully one of the DSO gurus will jump in here with some proven advice!
Al.