Question : 5axis simultaneous setup in LCNC

The Axis Preview assumes that the centre of rotation is the G-code origin. There is no way to tell it differently, so it is rarely accurate with 5 axis G-code.
It can still do the job of giving you something to look at to spot rapids through the stock and that sort of thing.
It isn't easy to think of a good way to inform the preview of the true rotation centre. Possibly it could be done with a "magic comment" in the G-code.

The Vismach / Impeller sample is TCP G-code running in a TCP Kinematics.

TCP means that the C-code XYZ is the position of the tool tip in XYZ space, and the tool ABC is the angle that the tool axis makes to the XYZ coordinate system So it is tool position and too tilt.

In non-TCP G-code the XYZ position looks very much the same, but it is axis position commands (with a trivkins machine this is pretty much exactly the same thing) but the ABC commands move the tilt axes directly.

It is probably easier to think of the difference by considering a robot arm. You can imagine programming the tool tip position either as an XYZ position + orientation in space, or program each joint of the robot individually. And in the latter case the limitations of G-code mean that you would need to still use XYZ ABC to program the joint positions regardless of the fact that they don't directly map to XYZ cartesian positions in space.

TCP G-code is a lot more portable, One program should work in all machines.
Joint-based G-code will only ever work in the machine it was created for, and furthermore the CAM system needs an exact internal model of the machine in order to make the G-code.