The "e above hull" field is incorrect for gaseous elements like H2, F2, etc. It always says "0" regardless of the energy of the structure. Some very unstable structures show e_above_hull=0, making them difficult to filter out.
This issue is originated from pymatgen Gascorrection class. A follow-up discussion can be found from https://github.com/materialsproject/pymatgen/issues/636 .
Note: N2 phases still all show ehull=0