Adapt EdgeTransformation operator to Tjoint.

[PaulineVidal]

For the computation of the equivalent coordinate, does not consider the T-joint case. It also affects the other operators.

Currently we deal with logical domains whose coordinates don't especially fit together so we rescale the coordinates:

• Eg: [a1, b1] sticks with [a2, b2] so a coordinate c1 on the patch 2 is computed with c2 = (c1 - a1) / (b1 - a1) * (b2 - a2).

 a1 || a2
 ||
 ||
 b1 || b2

But it doesnt work with T-joint.

 a1 || a2
 ||
 ||
 b1 || 
 |
 |b2

One solution is to consider that we always have (value of a1 = value of a2) and (value of b1 = value of b2). Which will be the most frequent case.

Another solution, more general, is to add the mapping(s) to the physical domain as input and work with it. It might be pretty difficult to identify if a coordinate is before or after the T-joint.