Sub-orchestrations cannot reuse existing instance IDs

[cgillum]

As of v1.1.0, the current implementation of sub-orchestrations doesn't support reusing instance IDs.

If you try to schedule an orchestration using an instance ID of some previously completed orchestration, you will see the following message in the logs:

2023-02-01T18:06:39.6352615-08:00 [DurableTask.SqlServer] XYZ: Discarding [ExecutionStarted#0]: Target is in the Completed state.

The issue is specific to the MSSQL backend. The same scenario works fine with the Azure Storage backend.