Bug with explicit energy coefficient

[CINTROINI]

The generalization of the compute methods for the SLOTH coefficients leads to an incorrect calculation of the second derivative of the explicit free energy.

[CINTROINI]

@tbarani @ascapini @niamorelreillet

[CINTROINI]

By default, when a SLOTH coefficient is explicit, its first and second derivatives are zero. However, for energy-type coefficients, such as the free energy and the interpolation function whose first derivatives are directly implied in the variational formulation, the first derivative must be nonzero and well defined in terms of the variables. The compute_gradient_energy_coefficient method enables this.

[CINTROINI]

Modifications are done in :

MeltingBaseNLFormIntegrator.tpp
CahnHilliardNLFormIntegrator.tpp
AllenCahnNLFormIntegrator.tpp
BlockAllenCahnNLFormIntegrator.tpp