Controller/InteractionClient Process optimizations and refactoring

[Apollon77]

This PR collects some things that I could not split better. Ithink it might be easier to review the single Commits because they for themself are simpler.

In fact the PR does:

• Cleanup in ChannelManager (especially centrallizing the Channelkey generation. It also adds "removeChannel" which is needed when InteractionCLient is reconnecting
• Create and throw own typed error for Retransmission Reached case to allow more easy catching
• Make timeout to discover a device configurable - during commissioning we usually wait 5s, for resuming a session (preparation for persistence!) we should allow to wait longer, so I made the timeout configurable
• Add session-Reconnect logic to "withMessanger" for Interactionclient to renew session - If we have a running connection to a device in controller case and the device gets restarted then the current session becomes invalid. Apple/Google/Amazon are opening a new session when this happens so we simply do the same. The "withController" method which is used gets enhances by a method that also can handle a reconnect if the Retransmission-Reached Error is thrown for the send command.
• The interactionClient gets an optional "reconnect" callback for this case to have the logic itself in MatterController where the rest of the connection login is owned. I think thisan elegant way to keep the topics selarated - if you have a better one just tell.
• The InteractionClient now also gets a method to "resume" a session and because we do something like that also when doing CASE pairing I refactored to code that the "connect" method takes care of that. It will all be used later when persistence is there - then the code can know if we have are commissioned or not and depending on this enter commissioning or just do a session resume/reconnect. So this is here preparation

[Apollon77]

@mfucci Ready for re-review ... but also see comment above

[Apollon77]

Reintroduced in matter.js