You could always devise a clamp system from the swing arm to the bar if not happy welding.
You'd want the pillow bearing as close to the swing arm as possible to ensure all upward force was transmitted to the twisting of the ARB rather than trying to bend it. The cast ones are very tough (I wouldn't use a stamped one) and ought to withstand any such force IMO. They also come in varying degrees of extra strength though I doubt they would be required. You can also run the arm through two or more pillow bearings should you wish. I suspect that would be extra weight for no gain. The fixed mount required could again be a bit of box drilled and slipped over the end and welded then bolted to the chassis. The idea would be to keep the ARB horizontal.
Round enough? - no idea at this stage but almost every cheap bike I've ever bought has had more slop in it's head tube bearing than I'd expect to find here. I have some shim steel so can add to the thickness to a degree and the bar must have been manufactured to some degree of roundness at the foundry. Many Pillow bearings have grub screws which would nicley take up a minor bit of play. As to whether a 12mm ARB really is 12mm or is actually 1/2" but someone didn't think the difference mattered I shall find out the hard way.
Zero camber changes with swing arms. It's swing axles that have that where the pivot is central between the wheels and the whole half axle pivots around that. Trailing swing arms are one of the best systems you can employ here. I can't think of any better system for the job in terms of geometry and weight. Every system that moves has some issue but with a 10" swing arm it's utterly ignorable. There's just over 1mm of fore and aft movement of the wheel as the wheel proscribes 2" or so of it's arc (assuming a swing arm largely parallel to the road) but there's absolutely no chance of that being any issue at all.
Bought
this one. Be aware that different variants of the same car may use different thickness bars though.
The longer the swing arm the better from the point of view of the arc it travels. Think of a ridiculous 1" swing arm then move the wheel axle 2" around that - it's doing the best part of a circle. Now expand that to the 10" I suggest and that 2" movement around the arc is almost in a straight line. Now from the point of view of leverage against the ARB, the longer the swing arm the greater the leverage and the fatter and/or shorter the ARB you need. Going longer on the arms will make it softer with the same ARB and shorter will be stiffer. In other words you really want a long (10" or so) arm for the geometry of the suspension if you have room and you then work out the correct thickness and length of ARB to suit. If space is a major issue I'd suggest you could go as low as 6" and then use a thinner and/or longer ARB to suit.
The length of the ARB is vital to it's effective spring rate. A longer piece will give softer suspension. It's exactly like a coil spring. If you take a biro coil spring and feel how easy it is to compress 1cm then cut it in half and to get that same 1cm squash you need twice the effort. Longer and/or thinner = softer. Shorter and/or fatter = harder. If you play around with different values in the
rate calculator you'll see that just adding or removing 1" to the swing arm or the ARB can have a sizeable effect.
There is a downside to welding the swing arm and the fixed mount to the ARB in that the bearing can not be removed for replacement. I'd get a pillow bearing with a grease nipple so it can be kept lubed and should replacement ever be necessary I'd grind the fixed mount off then weld a new one back on.