It seems that the LRA is correctly closed but the /lra-coordinator still reports it in Active status.
LRA forget calls for nested completed participants are not repeated and do not close the LRA
Nested participant AfterLRA call in complete scenario should be postponed until the parent LRA completes/compensates
Failed LRA is not kept in FailedTo* state in the object store
After LRA notifications can keep the LRA in Active state