Dear Users:
1. There was a bug in the calculation of the design orbit length
in a BEND with F1. The shift of the path length due to F1 is
-(phi0 f)^2/l/24 Sin[phi0(1-E1-E2)/2]/Sin[phi0/2]
but the term Sin[phi0(1-E1-E2)/2]/Sin[phi0/2] has been omitted
in the calculation of the path length. This effect was big in
the case of a rectangular bend, where the shift should be zero.
This error was only in the calculation of the geometry,
not in the tracking, etc. In the case of KEKB LER, this error
is about 1 mm out of 3 km circumference.