LEAmDNS_Fixes_1_3

[LaborEtArs]

Re-Fix 'avoid multiple initialization' code in LEAmDNSResponder::begin()

[devyte]

From gitter discussion:
@LaborEtArs with the UdpContext check, isn't the new hostname passed as arg to begin ignored?
Should the begin() implementation look something like this (unoptimized for the logic tree)?:

bool MDNSResponder::begin(const char* p_pcHostname) {

    bool    bResult = (0 != m_pcHostname);

    if (0 == m_pUDPContext) {
        if (_setHostname(p_pcHostname)) {

            m_GotIPHandler = WiFi.onStationModeGotIP([this](const WiFiEventStationModeGotIP& pEvent) {
                (void) pEvent;
                _restart();
            });

            m_DisconnectedHandler = WiFi.onStationModeDisconnected([this](const WiFiEventStationModeDisconnected& pEvent) {
                (void) pEvent;
                _restart();
            });

            bResult = _restart();
        }
    } else {
        if (_setHostname(p_pcHostname)) {
            bResult = _restart();
        }
        DEBUG_EX_INFO(DEBUG_OUTPUT.printf_P(PSTR("[MDNSResponder] begin: Ignoring multiple calls (Ignored host domain: '%s')!\n"), (p_pcHostname ?: "-")););
    }
    DEBUG_EX_ERR(if (!bResult) { DEBUG_OUTPUT.printf_P(PSTR("[MDNSResponder] begin: FAILED for '%s'!\n"), (p_pcHostname ?: "-")); } );
    return bResult;
}

Is the use of bool bResult = (0 != m_pcHostname); correct?

[LaborEtArs]

@devyte : You're right, the code was TOO 'quick and dirty', but I don't like your proposed code too, When 'begin()' was called once, another call to begin() should fail. At least it shouldn't change something silently (as it would, if OTA will be initialized after mDNS by the user). Unfortunately I can't change the code here (webeditor?, maybe only for branch owners? Maybe not with IE...?)...
The initial 'bool bResult = (0 != m_pcHostname);' should be 'bool bResult = (0 != m_pUDPContext);' if multiple calls should lead to 'false' or simple 'bool bResult = false;' (as I would prefer).
In addition the secend log should to moved to the end of the 'if-clause' to avoid misleading logs.

[LaborEtArs]

Updated (and included the 'schedule_function(std::bind(&MDNSResponder::_restart, this));' to avoid calling _restart() in SYS context.