As documented, the downtime of a child service does not affect its parent, unless the downtime is configured in parent as well, but it leads to a wrong SLA calculation.
If a parent has two children (A & B) and, for example, during the maintenance time of child A the child B gets down and up, the parent service cannot take both events into account.
If the maintenance period of child A is configured in parent, the event of child B is ignored in parent calculation.
if maintenance period of child A is not configured in parent, the parent calculus is obviously wrong because it takes the maintenance of child A as a unavailability period of A.
In both cases the SLA calculus is wrong.
In scenarios like the above the maintenance in one of its children should affect parent without registering that maintenance on parent itself. Maybe a flag or option should be create in further versions.