[example][bugfix] Change sign of flux in poisson-mfem

Starting from the Poisson equation -\Delta u = f and setting \sigma = \nabla u (or \sigma - \nabla u =0) we obtain \nabla \cdot \sigma = -f. Using partial integration we obtain the weak form \int \sigma \cdot \tau + u \nabla \cdot \tau= 0 of the flux equation. Thus the combined weak formulations reads:

\int \sigma \cdot \tau + u \nabla \cdot \tau + (\nabla \cdot \sigma)v = \int -fv.

So far the example contained a minus sign in front of both flux-pressure terms. This corresponds to defining \sigma to be the negative flux \sigma = -\nabla u. Hence the essential boundary condition for \sigma was in fact a negative flux BC. Maybe this was on purpose but I cannot see any reason for this.

This MR changes the sign, such that the essential BC for \sigma corresponds to a Neumann BC for u.

It also modifies the boundary values and rhs of the example, because the previous ones where so small that one could not see the sign error.