Thermo groups regression test

[davidfarinajr]

trafficstars

Adding and/or modifying thermo groups in the database can have unintended consequences for molecules outside a training or test set. This problem often arises with "supergroups" (groups that include neighborhood dimensions > 1) and "double counting" a particular chemical moiety included in supergroups. Making an RMG regression test that compares RMG thermo library values vs GAV estimates (and flags molecules where GAV predictions are worse with PR compared to library values) could help identity these problems in PRs.