fix: downgrade MQTT QoS from 2 to 1 to prevent dropped messages (#33)

QoS 2 (exactly-once) causes receive_maximum_exceeded errors on EMQX v5
during traffic bursts, as the broker's max_awaiting_rel limit (100) gets
overwhelmed. QoS 1 (at-least-once) uses a simpler 2-step handshake with
no awaiting_rel state, eliminating this bottleneck entirely. For chat
messages, at-least-once delivery is sufficient.

Related: emqx/emqx#13399 (broker returns wrong Receive Maximum in CONNACK)

Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: ffelipelimao

httpclient/httpclient.go

@@ -105,7 +105,7 @@ func (mc *HttpClient) SendMessage(ctx context.Context, topic string, payload str
 		Topic:    topic,
 		Payload:  payload,
 		Retain:   retainBool,
-		Qos:      2,
+		Qos:      1,
 		ClientId: fmt.Sprintf("arkadiko-%s", uuid.NewV4().String()),
 	}
 

mqttclient/mqttclient.go

@@ -106,7 +106,7 @@ func (mc *MqttClient) PublishMessage(ctx context.Context, topic string, message
 	l.Debug("Publishing message to mqtt")
 
 	for i := 0; i < 3; i++ { // Retry up to 3 times
-		token := mc.MqttClient.WithContext(ctx).Publish(topic, 2, retained, message)
+		token := mc.MqttClient.WithContext(ctx).Publish(topic, 1, retained, message)
 		if token.WaitTimeout(mc.Timeout) {
 			l.Debug("message published to mqtt")
 			break