Quaternion convention

[colormotor]

Hi Jean,
Not a bug, but I was testing the cartesian impedance controller and working on extending it to include orientation, and noticed that the expected convention for quaternions is with scalar last, which differs from the one in roboticstoolbox/spatialmath (which I am using quite extensively) and Eigen (in API not memory, confusing!). I wander if it is worth having a configuration flag that determines this? Otherwise I think it is worth pointing this out visibly in the documentation/examples as it can actually lead to some unexpected outcomes:)