Add bond orders to the OpenFF Group1 Torsions set

[jthorton]

Due to changes in connectivity that can occur during optimisations and torsiondrives bond order is now calculated on the final geometries to ensure bonding is consistent between the input and final molecules. However, the OpenFF Group1 Torsions set did not have this information calculated. There are two options to fix this as discussed on the QCFractal call

Option 1

Redo the entire data set, however this is one of the larger ones and is not necessary.

Option 2

Redo the last gradient calculation (at the optimised geometry) of the optimisation and replace the current final gradient calculation with the new id number. This will allow the dataset to act like others when trying to collect the dataset to form a fitting target.

I think option 2 would definitely be best in this case, @dgasmith is there anything you need me to do to start this request? Also, we decided that this is a low priority task.