For connectors with a manual schedule (i.e. only triggered manually via the Fivetran API or UI), if a sync fails due to an error then the sync should always end in an explicit failure state to indicate it failed.

As it stands, sometimes syncs for connectors with a manual schedule end in a "RESCHEDULED" state when the sync fails, even though Fivetran isn't actually rescheduling the sync itself.  For example, one of our SFTP connectors with a manual schedule recently ended in a "RESCHEDULED" state with the reason "Rescheduling to recover quickly after failure..."