Include gravity direction as input for structural solvers

[ArturoMS13]

Right now gravity always acts on the z direction in SHARPy.

However, xbeam allows gravity to be applied in any direction. Thus, it seems that we have ready the most laborious part of this feature and we are just missing the interface.

[ArturoMS13]

Now the gravity direction is a property of the simulation and not a beam property.

It is defined through the "gravity_dir" setting in structural solvers.

Everything should be backwards compatible except from solvers calling "cbeam3_asbly_static". I think the call to this function was not right. To my knowledge, this only affect static multibody computations.

I do not run many computations with gravity so any test or feedback is welcome.