BUG: leak closure in snappyHexMesh does not run in parallel

Summary

Leak closure produces very different results in parallel.

Steps to reproduce

Run shm (e.g. with debugFlags (mesh)) non-parallel and parallel side by side. At some point the cell count starts to diverge, e.g.

Removing mesh beyond surface intersections
------------------------------------------
..
Keeping all cells containing inside points
Selected for keeping : 1876028 cells.

Example case

Not easy. Make tiny case with small gaps. Use e.g. hierarchical. Run parallel with debug flags. Run

mpirun -np 9 postProcess -func processorField -parallel

and look at where the outside connects to the inside.

What is the current bug behaviour?

Inside mesh leaks outside but not at gap but at interface between two different processors.

Environment information

• OpenFOAM version : develop