Transactional conclusion

[matthiasgeihs]

Currently, the conclude functionality has two writes:

1. Set the outcome.
2. Conclude the registered state.

Problem: There is a risk that the first transaction goes through, while the second fails. In this case we might end up in an inconsistent state.

We need to investigate further whether the current solution is problematic and if so, how to resolve the situation. Maybe using the transaction decorator might be a solution, as pointed out here.