As a 585 user I have a little bit different point of view on this.
On bit depth. If you look at the graphs the 2600 the actual delivered dynamic range is just about the same. At a gain of 100 where HGC mode kicks in on both cameras the well depth is the same. This is where you will probably use but cameras the most because it gives you the lowest read noise in combination with almost the full dynamic range. To get the full well depth on either camera you would have to shoot at a gain of zero but at that gain read noise really goes up. This would only make sense if you are in the habit of shooting really long subs.
Since both cameras are delivering basically the same pixel level performance and the pixels are the same size you can eliminate that part of the argument and just look at field of view with your optics, cost, complexity, processing time and data storage requirements.
On FOV, if you are going to be cropping the 2600 most of the time it really doesn’t make sense to me to carry the cost, processing time and overhead verses composing your shots properly with the smaller sensor in the first place. You could take the money saved and get a rotator which would help a lot. I would really take a hard look at the vast number of objects out there and the needed FOV. The vast majority of objects are rather small, especially when you’re talking galaxies.
As someone who has gotten great prints from a 585 I really don’t agree with the idea that you need the larger chip for printing when the resolution is the same for both sensors. Yes if you make two 16×20 prints of say M51 at the full FOV of each sensor, the 2600 would look cleaner and sharper but that’s only because it’s enlarged less. If you made M51 the same size in both prints there would be no difference.
On the Duo, I would avoid it. It’s larger and heavier and brings in the single point of failure possibility for both imaging and guiding. It’s one of those everything is great until it isn’t things.
One point that might be very important depending on local conditions is that the 2600 has an internal dew heater, the 533 doesn’t.
Overall, unless you are really stuck on having a wide FOV and absolutely need an internal dew heater, I think the 533 is a better choice. In fact, with the money saved I would strongly suggest putting that into a mono setup.