Scattering width is the result of a two-dimensional calculation, and is for a shape for which the cross-section does not change in the third dimension, such as an infinite cylinder. The scattering width for a shape made into a long 3D cylinder (or any 3D shape that is constant shape along its length) is roughly the peak RCS (at broadside to the straight edge) times the wavelength divided by twice the length squared. This doesn't take account of all of the effects of the ends of the shape (which an infinitely long shape or 2D shape wouldn't have). If it is not the same shape all the way along then it does not have a scattering width - like a cube doesn't have a radius!
Ten wavelengths should be ok - the longer the better, but very long cylinders will take a long time in CST.
Perhaps do it for 3 or 4 different lengths and see if it is changing much. The problem is RCS due to currents near the ends, and possible waves travelling up and down the length of the cylinder. These become less significant compared to the total RCS the longer the cylinder.