The issue is outlined in the comment of the linked BF.

Possible fixes:

1. Change naming and semantics of metrics. Example name changes could be from resolved to acknowledged/receivedResponse, and from received to recognized/mirrored. The metrics can be placed in more precise locations based on names, and the assertions can be reorganized or changed.
2. Add a test only metric to measure how many responses were not successful. The current received metric is only incremented if the responses were successful, so assertions can be changed to consider how many responses were successfully completed.