[ojacques]

Problem

After a network outage (WiFi down for ~1 hour), zeClock could not reconnect to the ZeDMD because:

1. The ZeDMD's IP changed (DHCP renewal) but zeClock kept retrying the stale IP
2. In WiFi mode, UDP sends don't fail when the device disappears — zeClock didn't detect the disconnection
3. The retry loop only tried WiFi when wifi_addr was set, never falling back to USB
4. NetworkManager didn't auto-reconnect the Pi's WiFi (missing autoconnect-retries setting)

Changes

ZeDMD backend (zeclock/backends/zedmd.py)

• Detailed connection logging: logs the method (USB/WiFi) and target host at each attempt
• HTTP health check: probes http://<zedmd-ip>/get_version every 10s to detect silent WiFi disconnections (since UDP is fire-and-forget)
• Health check starts after 10s grace period to avoid false positives on first frame

Clock retry loop (zeclock/clock.py)

• Alternates USB and WiFi on each retry (odd=USB, even=WiFi)
• Always re-resolves zedmd-wifi.local via mDNS before each WiFi attempt — handles IP changes transparently
• Persists new IP to zeclock.ini when mDNS returns a different address
• Same mDNS-first logic in the runtime reconnection path (when send_frame fails)
• Replaced emoji print() statements with proper logger calls

WiFi setup (zeclock/wifi_setup.py)

• Added autoconnect-retries=0 (infinite) to connection profiles created by the captive portal

Testing

• Deployed on RPi, unplugged ZeDMD — logs show alternating USB/WiFi retries with mDNS resolution
• Reconnection works when ZeDMD is plugged back in
• 676 tests pass