fix(websocket): auto-restart after abnormal local close (pong timeout)

[krasnyt]

Problem

With setAutoRestart(true) the client fails to reconnect when the miniserver stops responding to websocket pings (a "frozen" connection). The connection stays dead until the application is restarted manually, even though auto-restart is enabled.

Root cause

When the remote end stops answering pongs, the underlying Java-WebSocket ConnectionLostChecker closes the socket locally with code 1006 (ABNORMAL_CLOSE):

LoxoneWebsocketClient - Closed by local end because of 1006:
The connection was closed because the other endpoint did not respond with a pong in time.

In LoxoneWebsocketClient.onClose() the restart is gated behind remote == true:

if (remote && code != CloseFrame.NEVER_CONNECTED) {
    ws.autoRestart();
}

A clean remote close (1000) recovers fine, but a 1006 local close — the most common real "frozen connection" case — never schedules a restart.

Fix

Also restart on an abnormal local close (1006):

if (code != CloseFrame.NEVER_CONNECTED && (remote || code == CloseFrame.ABNORMAL_CLOSE)) {
    ws.autoRestart();
}

A deliberate LoxoneWebSocket.close() goes through closeBlocking() and produces a NORMAL (1000) local close, so intentional shutdowns are still excluded and won't trigger a restart.

Tests

LoxoneWebsocketClientTest now covers three cases:

• remote close → restart
• abnormal local close (1006) → restart
• deliberate local close (1000) → no restart

[krasnyt]

hi @jimirocks , would you have time to have a look? thanks!