The 2905mm measured with a tape was done at same time the chain was tensioned during Maslow’s measurement last week. Going from 2888.6 mm to 2902.4mm (13.8 mm) is a big improvement.
When I upgrade to the latest firmware and autocal again (to 8113.73 _encoderSteps), I’ll recheck the tape measurement. Some of it (maybe 1 mm) might be attributable to how I measure from left side of left sprocket to left side of right sprocket. The left sprocket has chain wrapped on it and I eyeball where it would be on the right sprocket if it also had chain on it. I’ll put a link on the right sprocket to see if that makes a difference. It won’t be 2.6mm difference, but it will be closer.