Telemetry and message queues gain a dedicated 'urgent' priority lane

What

• A batching queue's pending-count now also counts items sitting in a new urgent sub-queue.
• Telemetry batching gained a takeUrgent() method that drains the urgent queue up to batch-size/byte limits, separately from normal pending-event batching.
• The outbound event/retry queue now tracks a separate urgent list alongside pending, drains both before considering itself flushed, and tracks a backoffOwedUntil timestamp once draining finishes.
• The backpressure/withdraw controller also tracks a separate urgent queue: withdrawing filters both queues, and closing folds both into in-flight/undelivered accounting.
• The durable event uploader gained an urgent fast lane that jumps ahead of the normal batch queue, with its own retry counter and backoff window, so an urgent send won't get stuck behind a slow batch's retry delay; sends that immediately follow an urgent flush are tagged followsUrgent.
• The mailbox/outbound event queue's enqueue() gained an urgent option that pushes straight into the urgent array, ends any active backoff early, and drains immediately, bypassing normal batching.

Why

This gives certain events a way to skip past normal batching and backoff delays instead of waiting behind whatever else is queued, while still counting correctly toward pending totals and backpressure accounting.