I would personally go with choice A.) in the first question, simply because then if ever want to export a full 1920x1080 copy you won't need to scale everything up to do so.. So I'd recommend you work inside a 1920x1080 then export to your recommended frame size when you're finished editing.
For question number two if you import full HD footage into a NTSC (720x480) sequence then you most likely won't ever need to scale the HD footage, because it will already be HUGE compared to the sequence you're working in. However if you do scale the HD clip up to 200 percent it will indeed make your picutre quality lower.
If you work with full HD footage inside a full HD sequence then export to SD or another smaller frame size the quality will lower, because the frame size is smaller therefore less pixels, but this isn't because of the workflow it's just because when you scale stuff down therefore making the picture contain less pixels the quality will appear lower.
So either way you go when you're taking HD footage then making it smaller you're going to be lowering it's quality, simply because there won't be as many pixels.
While the output would be the same, working on a sequence in the native resolution of the original footage (i.e. 1920 x 1080) would be **much** easier on the computer and require a lot less rendering. It is always best to work in the native resolution, output a native master, and then output lower sizes from there.
Well, again, always best to work in native res and scale afterwards. As far as "retaining the quality" goes... an NTSC sequence has NTCS quality. Your HD footage would look like really nice NTSC footage but it certainly wouldn't hold up on an HD screen.
I usually shoot everything as high as I can (within reason) to leave room for reframing in post as I need to. My final output chages based on target destination.