Just to follow up on what Dan has written about the concentricity of the swingarm pivot and the bevel box sprocket; on my first chassis (late 2012) it was very close to concentric so very little change in belt tension with swingarm travel, the second chassis was much the same but with my current chassis there is a little bit off a gap between the centres so the tension does change as the swingarm moves. My chassis is a slightly odd hybrid of the later type with welded in front bracing but with the earlier, solid bevel box mounting. I suspect that the rubber mounting for the bevel box may increase the gap between the centres so there is a greater change in belt tension.
Incidentally, in bike chassis design the offset between the centres of the swingarm and sprocket centres is deliberately played with in both distance and height as it affects the way the back tyre finds traction under power. Various bikes have tried the concentric arrangement but it generally hasn't led to good handling though chain life might be good!
