Operational features round 1

[bmfmancini]

• feature: Add a fail-safe for partition maintenance: when the partition metadata for the syslog or syslog_removed table is invalid or incomplete, or a required partition cannot be created, all partition creation and retention pruning stops, the dMaxValue safety partition is never dropped, and the blocked condition with its remediation step is logged and shown on the Syslog Status page
• feature: Add bounded multi-day partition recovery: a new 'Partition Recovery Limit' setting caps how many missing partitions are created per table per poller run so large gaps heal over several cycles; retention pruning stays deferred until the future write horizon is restored, and each blocked run reports the remaining gap, the stop reason, and whether writes are accumulating in dMaxValue
• feature: Add collector health metrics to the Syslog Status tab: last received log timestamp, oldest unprocessed incoming message age, current incoming backlog, records processed in the latest run, and a warning when data looks stale or the backlog exceeds the new 'Collector Staleness' and 'Collector Backlog Threshold' settings; unavailable metrics read as Unavailable instead of invented values
• feature: Add per phase processing telemetry: the poller records start time, end time, duration, and processed count for partition maintenance, reference updates, removal rules, alert evaluation, incoming transfer, and report processing through the syslog_status table; the Syslog Status tab shows the latest phase timings and highlights the slowest phase
• feature: Add a 'Test rule' action to the Alert and Removal rule editors: a read-only preview compiled through the existing QueryBuilder and parameterized SQL paths returns the matching count plus a bounded sample of escaped messages, without saving, enabling, deleting, alerting, emailing, executing commands, or altering any data; the action respects realms and CSRF protection