Fix local push reconnection logic to prevent stalled reconnections

[Copilot]

Summary

The reconnection logic had a critical bug where failed reconnection attempts would not schedule subsequent retries, causing local push to permanently fail until app restart. Additionally, rapid status updates could schedule duplicate timers with incorrect backoff timing.

Root cause: scheduleReconnection() was only called when a server was first added to disconnectedServers. After the first reconnection attempt failed, subsequent status checks found the server already tracked and did nothing, breaking the reconnection chain.

Changes:

• Renamed scheduleReconnection() → scheduleReconnectionIfNeeded() with idempotency guards
• Always call scheduleReconnectionIfNeeded() for unavailable states (not just initial disconnection)
• Guard against duplicate timers: skip scheduling if timer already active
• Clear timer reference when fired to enable next attempt scheduling

Flow before fix:

case .unavailable:
    if !disconnectedServers.contains(server.identifier) {
        disconnectedServers.insert(server.identifier)
        scheduleReconnection()  // Only called once
    } else {
        // Subsequent checks do nothing - reconnection stops!
    }

Flow after fix:

case .unavailable:
    if !disconnectedServers.contains(server.identifier) {
        disconnectedServers.insert(server.identifier)
    }
    // Always attempt scheduling - idempotent guards prevent duplicates
    scheduleReconnectionIfNeeded()

Screenshots

N/A - Internal networking logic change

Link to pull request in Documentation repository

Documentation: home-assistant/companion.home-assistant#

Any other notes

This PR merges into #4113 and addresses feedback from review comment #4113 (comment).

---

✨ Let Copilot coding agent set things up for you — coding agent works faster and does higher quality work when set up for your repo.