feat(esp32): #83 brownout defense-in-depth (PHY TX backoff + connectivity log) - #94

…vity log)

Two firmware levers for the node 35 brownout crash-loop at phy_init. Neither
touches the initial phy_init calibration PEAK that trips the reset (only the bulk
cap / custom PCB #11 does) -- these are belt-and-suspenders.

1. CONFIG_ESP_PHY_REDUCE_TX_POWER=y (sdkconfig.defaults, grouped with the brownout
   block). IDF-native: esp_phy_reduce_tx_power() forces the lowest PHY TX power on
   the boot immediately after ESP_RST_BROWNOUT (phy_init.c gates it on exactly that
   reset reason), trimming current on the retry boot. Post-brownout backoff, NOT a
   peak cut.

2. Wi-Fi connectivity-transition logger (app_main.cpp). Originally scoped as a
   runtime reconnect backoff via ConnectivityMgr().SetWiFiStationReconnectInterval,
   but esp-matter's ESP32 ConnectivityManagerImpl DECLARES _SetWiFiStationReconnect-
   Interval without defining it (only _Get links; the interval is a fixed compile-
   time 5s), so that call does not link on this platform. CHIP already auto-
   reconnects on its 5s timer, and the field "never reconnects" symptom is the
   brownout reboot loop (pre-CHIP) anyway. So this ships as an observer that
   timestamps Lost/Established transitions -- lets a real Wi-Fi drop be told apart
   from a brownout reboot when triaging a node. Observer-only; touches no CHIP state.

Bumps PROJECT_VER 1.1.7 -> 1.1.8 (int 10108); sdkconfig NUMBER/STRING synced.
Verified: builds + links clean (debug flavour, 15% flash free). Not yet flashed to
node 35. See #83.

Assisted-by: AI