v4 June 2019 Training reaction fixing hackathon

[ajocher]

trafficstars

Hackathon had two objectives:

1. Remove unreasonable high rates that prevented us from setting a bounds on rates within each family for filtering. If the literature source is considered reliable but maybe the available species thermo data is not, the training reaction has been moved to the new repository 'training_insufficient' within RMG-database/input/kinetics/families/<family name>/training_insufficient.
2. Remove unnecessary duplicate reactions from the database.

Emergent considerations: Refitting GAV estimate training reactions to stay within collision limit. Moving training reactions with small temperature ranges to RMG-database/input/kinetics/families/<family name>/training_insufficient

[ajocher]

• Label fixing commits squashed. All fixes in a family are now in a single commit.

• The commits adding the jupyter notebooks are removed, because we think we don't need the notebooks in the future, if the notebook to generate training reactions is updated with a collision limit check.

• Reactions that have been removed from reactions.py are now stored in a new repository training_insufficient for further study/confirmation or thermo calculation/selection before they can be put back into reactions.py.

• Removed training reactions for R_Addition_MultipleBond are now in the new repository and each training reaction has a reason for removal documented in the comment. Hope this will address @mliu49 comments above?

• Haven't gone through H_Abstraction.