Add option on re-assmble to allow publishes if some messages in the subset go into an error state

[jhorbulyk]

Other Feature Request

Description

Currently, the re-assemble action will not publish a re-grouped message if a single child message of the group goes to an error state. In some use cases, this might be a problem. As a result, there should be some mechanism (perhaps time based) which would see a newly grouped batch become published if it can be assumed that a child message has errored or otherwise become stuck.

Definition of Done

Exact params of solution TBD.

[jhorbulyk]

Implemented in #72