Improve same media playback completion handler

[defagos]

See https://github.com/SRGSSR/playsrg-ios/issues/154.

[defagos]

As discussed with @oiledCode, and in addition to the changes proposed in the Play SRG issue above, it is likely we should also call the prepare to play completion handler even if in some cases we do nothing. This would provide for a more consistent behavior.

This change has been made on a feature/completion-consistency for testing purposes. A corresponding unit test should also be written.

[pyby]

A unit test has been added on feature/completion-consistency.

[defagos]

Not sure this fix was a good idea, see https://appcenter.ms/orgs/RTS-Organization/apps/Play-RTS-1/crashes/errors/4277850553u/overview.

If we exit early and call the completion handler, this means the -play which might be called from it might not find any content URL, prepare playback again, thus ending in an infinite recursion. Relevant code:

• Raw play
• Play with parameters
• Early completion.

Some ideas:

• In the play with parameters implementation ) call -play on the media player controller directly. Problem is that the Letterbox controller play must be able to handle the same cases as today, and that clients of our Letterbox controller API could still do what we currently do. So we would not fix the issue.
• Only call the completion handler early if there is a content URL. This seems hacky and is inconsistent.
• Remove early completion handler call, i.e. revert this fix, but we must check with @oiledCode first whether this behavior is still being used in their code.

A proper fix would require #227 to be implemented anyway.

[defagos]

We had a rare crash in production which seems to confirm this intuition.

This crash is not related to an other infinite loop which was fixed at the SRG Media Player level.