Wrong number of neighbours

Currently, we are assuming that c->h_max is always the largest h but this is not the case. Therefore the optimized version of the iact is missing a few particles (less than one particle over 1000).

We can solve it with two different ways either we ensure to have a meaningful c->h_max or we do not break the for loop when reaching c->h_max (https://gitlab.cosma.dur.ac.uk/swift/swiftsim/blob/master/src/runner_doiact.h#L1004 and the one below)