I generally dump the whole set through audiogate as one track then if I need to normalize I take it to my DAW and do it there normalizing to -0.5dB over the whole set. Definitely don't normalize per song(you probably know that already). This isn't to say the normalize wouldn't work in audiogate, it just seems primitive, I have greater flexibility on what level to normalize to in my DAW. I suppose I should just normalize a test track sometime through audiogate to see what it does, the "normalize to average volume" just kinda scares me. As far as editing, I find that much easier to do in a DAW where I can see the waveform and work with multiple track overlays for patching etc. I then use CDWave to track so there are no SBE's and use it to flac as well at level 8, I believe audiogate will use system default which is level 5, audiogate is also not SBE aware so you will have sector boundary errors if you use it for tracking without fixing SBE's with TLH or another such tool.
my usual processing chain for DSD:
Korg MR-1(DSD of whatever flavor) > audiogate(decimate to wav/dither[Korg Aqua]/resample) > DAW(normalize/edits if needed) > CDWave(SBE friendly tracking/flac'ing) > FLAC(Level 8[or whatever other format, be it mp3,ogg,whatnot])
If the audiogate normalize had more functionality I would certainly use it and take the DAW out of the equation for anything I didn't need to do edits on. Not sure if this helped or not, at least it's of an example of one way to go about it.
J