Real time client memory leaks fixes

[lukasz-szyszkowski]

This change was originally made by @mikepulaski in ably-cocoa fork (https://github.com/tupleapp/ably-cocoa/tree/gcd-leaks).

[QuintinWillison]

Looks solid. Any ideas why checks aren't passing?

[lukasz-szyszkowski]

Looks solid. Any ideas why checks aren't passing?

well, it looks weird ... because sometimes they pass and sometimes not. I'll re-run all jobs an we will see if they'll pass or not

[lukasz-szyszkowski]

I double-checked the failing GCD iOS test and there was an issue with this https://forums.swift.org/t/retain-count-set-to-2-and-no-deinit-called/43571/4
and finally, I changed it to this https://www.swiftbysundell.com/articles/using-unit-tests-to-identify-avoid-memory-leaks-in-swift/#closures

commit here: 0f8e29f

[maratal]

I would mark each line with retain/release counter, such as "+1 to scheduledBlock, total = 2". Otherwise it's very frustrating to read this with a fresh eye.

why? In this solution, we don't need to check to retainCounter, this was in the previous solution and it was doubtful.

Just to give the next reader a quick idea on what's going on in this test.

[lukasz-szyszkowski]

I would mark each line with retain/release counter, such as "+1 to scheduledBlock, total = 2". Otherwise it's very frustrating to read this with a fresh eye.

why? In this solution, we don't need to check to retainCounter, this was in the previous solution and it was doubtful.

Just to give the next reader a quick idea on what's going on in this test.

here 6df222c