reduce2 complaining about too-high i_seq numbers in some models.

[russell-taylor]

When fixing other problems with 8zst.cif, warnings showed up about invalid i_seq numbers (higher than the maximum). These used to occur on other models that did not have the same problems (1aw9, 1ca1, 1cfz).

These problems are occurring (again?) on 1aw9. With 8zst, it was suspicious that the number of overages matches the number of atoms with missing info (GD atoms), but this may be coincidence.

We had i_seq problems after removing hydrogens (from Histidines), but I think we resolved that issue.

This is presumably also mis-labeling other atoms in the file, associating them with the incorrect atom information and so forth.