Release v3.22.0

Changed

• cratis chronicle diagnose supports --all-namespaces and --all-event-stores, with findings identified by event store and namespace. (#216)

Fixed

• Diagnostic sweeps report unavailable checks separately from detected problems and return a non-zero exit code for incomplete checks or quarantined observers. Watch mode returns the last sweep's result. (#216)
• Watch reports show health and unavailable checks first, mark unavailable checks as failures, and limit findings to three rows with guidance for viewing the complete report. Before the first sweep, every check row shows a neutral checking state and aggregation headers identify all selected stores and namespaces. (#216)
• Store and namespace discovery failures mark dependent diagnostic checks as unavailable. All-store diagnostics no longer prompt to change the default event store, and completed multi-scope event sequence checks point to per-scope tails. Empty logs are distinguished from failed tail queries, and successful scope tails remain visible in incomplete aggregated reports. Aggregated JSON and plain reports keep event sequence tails per scope; single-scope reports without aggregation no longer duplicate the tail row. (#216)
• Plain diagnostic output keeps values on one line, quotes and escapes field delimiters, and places scalar keys before repeated records. JSON observer totals count every returned observer, including unknown and quarantined states, in both aggregate and per-namespace results. (#216)