Populate `st_blobs` on restart in checkpoint/restart workloads in adapters

[ChristopherHogan]

This applies to any workload that runs multiple client applications against a single Hermes daemon.

The adapters maintain their own metadata, so when an adapter-intercepted app exits, it loses all info it was maintaining, even though the data may still be kept in a Hermes instance that was kept alive with HERMES_STOP_DAEMON=0. When the next adapter run starts, it needs to check the running Hermes instance to see if files it's trying to open or fopen already exist as Buckets. If so, it needs to get all the Blob names of that Bucket and fill the AdapterStat::st_blobs member with that information.