Draft: Slightly reorder skipping check of h interation

Currently the check for skipping particles because, e.g., they want to increase h to find neighbours but are already at h=h_max, happens after a debugging check that the bisection bounds are valid. So occasionally the bounds are found to be invalid even though it's already a situation where they aren't relevant for a particle that should be skipped.

Here just moved the skipping check to be slightly earlier, immediately after calculating the newton-raphson f that enables the check.