Improve PINGREQ <--> PINGRESP reliability during network change

[pglombardo]

🚀 Feature Request

With TCP being a stateful connection, for any PINGREQ sent, we expect a PINGRESP to be returned. This logic breaks when there is a network change between REQ & RESP.

The client should allow some ping responses to be discarded and instead retry the ping again.

This issue was pointed out in #508.

The code in question: MqttPingHandler.java

🔈 Motivation

🛰 Alternatives

📎 Additional context