Fix keep alive checking interval

[muneebahmed10]

Description:
As pointed out in https://forums.freertos.org/t/keepalivetimeout-code-in-coremqtt/12957/3, when a ping request is sent from MQTT_ProcessLoop(), it is only checked for a response after the keep alive interval has elapsed again. Since the broker would have disconnected the client by that time anyway if the PINGREQ was not received, the response ought to be checked earlier, as dictated by MQTT_PINGRESP_TIMEOUT_MS.

Two solutions for this issue are:

1. Update handleKeepAlive to check based on pingReqSendTimeMs and not lastPacketTimeMs.
2. Update MQTT_Ping() to only update pingReqSendTimeMs and not lastPacketTimeMs.

The latter solution would mean that sending other packets, e.g. a PUBLISH, would reset the interval before the PINGRESP is checked, while in the former it would not. However, the latter solution results in a smaller binary when optimizing with -O1. This PR implements the former solution, and leaves discussion for the tradeoffs for this review

[muneebahmed10]

I thought 5 seconds was a more reasonable default for this, for slower network conditions. From just testing locally with QEMU, it takes > 1 second for a CONNACK to be received, and sometimes with device advisor, the CONNACK takes 5-10 seconds to arrive. Since this bugfix will make this macro actually be used, we might be getting more bug reports from customers if we don't update this value now too.