Handling reconnect scenarios properly when socket is hung

[markshiz]

In situations where the socket connection is lost via a SO_ERROR, even though reconnect is enabled, the socket manager hangs the socket until a manual disconnect. Symptom would be: the app would output something like the following:

2019-01-09 12:15:49.005630+0000 App [13746:3543026] [] nw_socket_handle_socket_event [C10.1:1] Socket SO_ERROR [61: Connection refused]
2019-01-09 12:15:49.016640+0000 App [13746:3543026] [] nw_socket_handle_socket_event [C10.2:1] Socket SO_ERROR [61: Connection refused]

Socket-IO would be logging the retry requests over and over again with the following log:

Tried connecting an already active socket

In this case, the active socket is hung, so a retry would never be attempted. This condition can be simulated by running an app on a simulator while turning on Network Link Conditioner with 100% packet loss profile.

This fix simply recycles the engine in the case that a connecting status has 1 or more retry attempts. This seems to resolve the issue.

[WestFlow127]

@nuclearace this PR fixes the issue for 1415#

This bug is critical. Can we please get this merged?

[nuclearace]

@WestFlow127 Would you be able to test on development if I merge? I don't have capacity this week

[WestFlow127]

@nuclearace Yes, I can do that.

[WestFlow127]

Thank you! @nuclearace

[ricardopereira]

@WestFlow127 did it passed your testing?
Also, any idea when this fix will be released? Thanks

[WestFlow127]

@ricardopereira Yes, I have just recently fully tested this against my companies socket server.
My testing passed, and worked well. I have tagged the repo owner for requesting a merge of this PR in master:
#1464

[ricardopereira]

I also tested it. I was facing issues with reconnections and this PR fixes it. Thank you!