Description
Issue
When using DT.AzureStorage v1.8.4 or earlier to start orchestrations that get picked up by a worker running DT.AzureStorage v1.8.5 or later, the orchestrations will get permanently stuck in a "Pending" state.
Root Cause
This is a regression caused by the orchestration de-duplication start PR: #528. The de-dupe detection assumes that newly started orchestrations will have an ExecutionId field in the Instances table, which is true for v1.8.5+ clients. However, if this field doesn't exist, the de-dupe logic incorrectly flags the start message as a duplicate and discards it.
Workaround
When upgrading from DT.AzureStorage v1.8.4 or earlier to a newer version of DT.AzureStorage, you must either ensure all clients and workers are upgraded together or ensure that all clients are upgraded BEFORE any workers get upgraded.