MIELHVAC fix(miel_hvac): memory leak, remote temp logic, code deduplication

[grzegorz914]

Description:

This PR fixes three bugs (one causing a spurious serial write on every boot), corrects a wrong lookup table in HVACSetPurify, removes ~80 lines of duplicated JSON-building code, and cleans up misleading comments.

---

Bug fixes

Memory leak in miel_hvac_pre_init() (#1)

When sc_serial->begin() failed the code executed goto del, which deleted the serial object but never reached free(sc) — the free: label existed but was never jumped to. sc leaked on every failed initialisation. Fixed by replacing the goto chain with an explicit delete + free + return.

// before
del:
    delete sc->sc_serial;
free:           // dead label — never reached
    free(sc);

// after
    delete sc->sc_serial;
    free(sc);
    return;

remotetemp_clear semantics were inverted (#2)

The variable was named remotetemp_clear but acted as an "active" flag: setting it to true meant the sensor was active, yet it also armed the auto-clear timer and was published to MQTT as "on". Renamed to remotetemp_active with corrected semantics throughout (true = remote temperature override is active, false = no override).

Spurious CLR frame sent on every boot (#3)

The auto-clear condition included remotetemp_last_call_time == 0, and the old remotetemp_clear was initialised to true. Combined, this caused a remote-temperature CLR packet to be sent to the HVAC unit on every boot before any sensor had registered a value. Fixed by initialising remotetemp_active = false and removing the == 0 branch from the timer check.

// before
if (remotetemp_clear &&
    (remotetemp_last_call_time == 0 ||          // fires immediately on boot
     millis() - remotetemp_last_call_time > remotetemp_auto_clear_time))

// after
if (remotetemp_active &&
    millis() - remotetemp_last_call_time > remotetemp_auto_clear_time)

---

Correctness fixes

HVACSetPurify used the wrong lookup table (#5)

miel_hvac_cmnd_setpurify() was passing miel_hvac_airdirection_map to miel_hvac_map_byname(), so the command accepted indirect/direct/even/off instead of on/off. Restored and enabled miel_hvac_purify_map with values {0x00, "off"} / {0x08, "on"} (verify 0x08 against your hardware model).

Misleading hex comments in miel_hvac_cmnd_setairdirection() (#9)

The switch cases had comments like // 0x80, // 0xaa which referred to widevane wire values, not to the airdirection #define values (0x00–0x03). Removed the misleading annotations and replaced them with a single explanatory comment describing the i-see sensor / widevane relationship.

---

Refactoring

Deduplicated settings JSON builder (#6)

miel_hvac_publish_settings() and miel_hvac_sensor() contained ~80 lines of near-identical code building the settings JSON payload. Extracted into a new static helper miel_hvac_append_settings_json(). Both callers now invoke the helper; any future field additions require a single change.

---

Minor improvements

• htons() usage on flags documented — added a comment explaining why network byte-order conversion is used (wire protocol is big-endian; ESP8266/ESP32 is little-endian).
• operationtime documented as unverified — added a comment clarifying the 3-byte big-endian layout and noting that the unit (presumed minutes) has not been confirmed against hardware.
• Local char buffers in timer/status JSON consolidated — four separate char[33] declarations replaced with a single reused buf[33].

---

Testing

Tested on ESP32-S3 with a Mitsubishi MSZ-LN series unit. Verified:

• MiElHVAC sensor object populated within a few seconds of boot
• no CLR packet on boot
• HVACRemoteTemp 22.5 sets and auto-clears correctly after the configured timeout
• HVACSetPurify on/off accepted and transmitted correctly
• HVACSettings MQTT topic content unchanged

Checklist:

• The pull request is done against the latest development branch
• Only relevant files were touched
• Only one feature/fix was added per PR and the code change compiles without warnings
• The code change is tested and works with Tasmota core ESP8266 V.2.7.8
• The code change is tested and works with Tasmota core ESP32 V.3.1.11 / V.3.3.8
• I accept the CLA.

NOTE: The code change must pass CI tests. Your PR cannot be merged unless tests pass