Coordinate one Scheduled Reopen

Parent

#412

What to build

Give unexpected Transport loss one coordinated delayed Reopen. Recovery requests share the existing schedule without postponing it, and each recovery attempt has a clear success or failure outcome.

Acceptance criteria

• Unexpected loss of the currently owned Transport schedules exactly one delayed Reopen.
• Repeated delayed-recovery requests share the existing schedule and do not reset its deadline.
• A scheduled Reopen is consumed before its Connection attempt begins; Scheduled Reopen and active attempt never coexist.
• A failed recovery attempt creates exactly one fresh Scheduled Reopen at the configured fixed delay.
• Successful recovery leaves no pending schedule and retains the established Transport.
• Stale or duplicate Transport callbacks cannot create recovery intent or duplicate timers.
• Recovery lifecycle events occur once per authoritative attempt and joined recovery requests do not multiply them.
• Deterministic timing tests cover scheduling, repeated requests, failure, success, and stale callbacks.

Blocked by

• #413

Read parent issue #412 and blocker #413 first to understand the recovery lifecycle and existing scheduling behavior. Then identify the Transport loss and Connection attempt entry points and add deterministic timing tests for scheduling, repeated requests, failure, success, and stale callbacks. Done means all listed acceptance criteria hold without duplicate schedules or lifecycle events.