I just noticed the following thing (and I assume other people will eventually find it odd too, e.g. #868)

The value for the Hartree changed slightly, but significantly

27.211'385'05(60) eV CODATA 2010
27.211'386'245'988(53) eV CODATA 2018

This is enough that values like SCF energies become difficult to find, when one compares the value cclib parses with the output file.