[GSPH] h is not guaranteed to be converged before continuing

[tdavidcl]

In the SPH solver we loop until all particles have a converged h, not in GSPH. I would expect that this impact the solution quality