handle_invalid_molecules branch

[EBjerrum]

Hi @c-feldmann , thanks for the pull request and contribution. There were some issues with the pytests, and I merged to a branch to examine it. That of course made the pull request disappear (doh!), so you maybe don't see my comments. However, the code's not ready for merge to main, and just for clarification and to avoid double work and code conflicts, is it something you intend to work further on? Or should I simply take what I can use from the code and implement sooner or later?

[EBjerrum]

@enricogandini Hi, do you have time for a chat? I've got an interesting pull request to handle invalid mols (which is very nice for deployment and productionizing models). However, there are some interference with the pandas output, and I would like to discuss it with you, since you are the expert on that part of the code.