I'm not sure what to make of this either.. are you sure you executed px_d8 without the subchannel command? so 'px_d8 d 0 0'.. if not, there's a chance the sector offset is reported wrongly.. Also, you have to disable subchannel reading in cdtool when reading the sector.
So if both px_d8 and cdreader were used without subchannel reading and it still gives you the same scrambled header in sector -1, the maths of 270+588-98 (I suppose you're sure the read offset is correct?) = +760 is correct
And is 1.00 also the gap that perfectrip reports?
And you should notice the sync/header in the track02 pregap output.. it indicates the same offset of +760 (if you're sure about the +588) for all drives! So it's best to use this value for dumping.. I have no idea why the isobuster offset is this big, but it may be a mastering error.. have you tried cdtool instead to read the same sector? or try going back -73 first and then one more (I also don't think that d8 will ever give you a write offset that isn't a full number in samples, but let's wait what themabus has to say on this)
The only time d8 and the classic method don't provide identical results is when the data track doesn't have a write offset (it will only show the read offset with d8) but the audio does... this time you see a sync/header in the track02 pregap scrambled data output which normally doesn't show, so for some reason you'll also have to use the sync position this time, like in the d8 method..
I just hope there aren't any other discs with the same problems.. the write offset for those might have been calculated wrongly already (if the scrambled data output has a sync/header it's best to use its position instead of the amount of scrambled data).