Release v19.29.0

Added

• ClearPartitionQuarantine operation on the Observers service (with ClearPartitionQuarantineResponse) clears the quarantine of a single failed partition, starts a new retry budget for it while keeping its attempt history, and can retry it immediately. Every refusal has its own outcome: not found, not quarantined, or the observer itself being quarantined. (#4407)
• ClearFailedPartitionQuarantineFor<TReactor>() in the .NET client clears and optionally retries a quarantined reactor partition.

Changed

• maxRetryAttempts now applies per retry budget, so a partition whose quarantine was cleared gets a full set of attempts again. Alert evaluation counts attempts against the current budget.