Fix for issue #4 , bug fix for set_position when hard limit is reached with a pseudo motor, improvement of Group state

This merge request tries to fix the problem demonstrated by issue #4 (closed), without introducing a new wait_start option to motor group move method. I think the problem really was a bad reporting of states in the motor group, hence this merge request comes with an improved state method, composing a state with sub-states of each motor, so now limits are properly taken into account in Group objects. It also fixes a bug that was found by Carole Clavel during tests of the QSys table for ID23-2, set_position was not reset when a real motor touched a hard limit when moving a pseudo motor. The commits also include tests, for all motor control related issues ; I think we should be more strict with testing, I propose to not accept any merge request related to motor control at least if there is no test with it.