Issue with the NF in Chemical library

[Francesco Casella]

Recent commits to the NF broke some models of the Chemical library, see
​report1 ​report2.

In the meantime, @marek applied a ​workaround to his library that ​reverted one issue, but I suggested him to undo it, so we can get the NF properly fixed instead, as this issue could affect other libraries as well.

[Per Östlund]

This had already been fixed before the ticket was opened, see ​report 1 and ​report 2. Some of the models that regressed still look worse than before my changes if you look at the coverage report for the library, but that's due to library changes made later.