investigate culprit for why residualMembers can get into inconsistent state

[ikbalkaya]

We should investigate why residualMembers can get into null when endSync is called.
Context :

Yes - you are right, I had a check and the access is protected. In this case, I think doing this might remedy crashing the app, but it means that we are probably covering up another important bug - As far as I can see the only place residualMembers is null can be null is where it's declared and in endSync method. In this case it means either endSync is invoked early (before startSync) or is invoked after another endSync which would be problematic. I will approve this to remedy the crash but I think we should create another issue to investigate this situation.

Originally posted by @ikbalkaya in #873 (comment)

[sync-by-unito]

➤ Automation for Jira commented:

The link to the corresponding Jira issue is https://ably.atlassian.net/browse/SDK-3150

[davyskiba]

Just to make sure we're all on the same page, the exception that we've seen is the following:
java.lang.NullPointerException: Attempt to invoke virtual method 'java.lang.Object io.ably.lib.types.PresenceMessage.clone()' on a null object reference.
The only place I've found where this could occur is here

I think it happens when there is a key in residualMembers that is not present in members, null is received from members, and then cloned.