[IRC] Reset the restPositions while fixed - make consistent StartPosition

[fredroy]

In 3instruments_collis, we can notice a slight bump at the beginning of the deployment of the catheter.
One of the reason is that the first liberated point (when the deployed length increases and need to "free" one DOF) when freed, first go its restPosition. And in this scene, the restPosition is set by the RegularGrid (and is a linear set line of points)

So to fix this, while "fixing" the undeployed tool, the restPosition is also set at startPosition.

EDIT: actually this PR is much more important than only fixing the "bump".
Now, the starting orientation of the beam is solely set by the startingPos data, and no more from the orientation (?) of the grid which was weird.

Illustration of the bump:
Before

After

And illustration of the shenanigans of the bug in the state position: (this is supposed to go along the Z axis)
Before

After

[bakpaul]

Doesn't this affect the internal forces of the beam ? The deformed state being compared to a wrong rest state ?

[hugtalbot]

Checking the regression raised the fact that the error on DoFs was significant. This has beem fixed along with a deeper bug on the definition of the initial orientation. Data startPoint is only used for the position, the orientation is given by the topology used. Now only the startPoint data matters and the PR is more breaking.