Memory Leak about aws-mqtt

[cgood92]

After looking through the mqtt library that this is extending, it looks like the first parameter of the super (the callback) is called under a few different scenarios. That's fine. What we should just do is that in createWebSocketStream, before you create a new Websocket, you should clean up the old websocket that was created.

[kmamykin]

I will take a look at this over the weekend. Feel free to submit a PR if you already have this fixed.
Thanks!

[cgood92]

I take it back. It was my fault. I had 2 issues:

1. I wasn't cleaning up event listeners
2. I misunderstood just how critical clientId uniqueness must be per client (I was thinking "per user", but that was wrong - per client is the correct terminology). That was pointed out in the README, but I misunderstood it.

So, I have made this MR with my 2 issues documented, in hopes to help other people not have this same problem: #6.

The solution I pointed out above does not seem to fix anything real, so I say we leave it as it is.

Thanks,

[RelaxLiu]

@cgood92 @kmamykin
After the net is broken, it is reconnected, but there is no data.

publish-subscribe

Subscribe to success

Data reception success

Off the net

Link offline

Reopen the network

After that, the link has been successful. Close, offlin, connection is successful, close, offline has been circulating, neither resubscribe nor cancel subscription.

Want to ask what you need to set up to continue the data transfer