During app fabric startup pipeline recovery recovers each pipeline multiple times, leading to launching overcounting

During app fabric startup we perform scan and recovery of active runs in ProgramNotificationSingleTopicSubscriberService#doStartUp. With topic sharding this topic now runs multiple times, equal to number of shards. This leads to multiple recover messages for each active run and overcounting or “starting” runs”.

We should move the code from ProgramNotificationSingleTopicSubscriberService to ProgramNotificationSubscriberService