V2: Introduce cache cap, backoff when retrying previously failed events

[kkostov]

This PR fixes #284 by capping the maximum number of events we keep in cache and introducing a back-off retry mechanism for previously failed batches.

• TelemetryManagerConfiguration has been extended to allow apps to configure this behaviour as needed:

  Property	Type	Default	Description
  cacheLimit	Int	10_000	Maximum events held in memory
  transmitInterval	TimeInterval	10	Seconds between transmission attempts
  maxBackoffInterval	TimeInterval	300	Upper bound for the exponential backoff delay

  let config = TelemetryManagerConfiguration(appID: "<your-app-id>", namespace: "com.example")
  config.cacheLimit = 5_000
  config.transmitInterval = 30  
  config.maxBackoffInterval = 600
  TelemetryDeck.initialize(config: config)

• Fixed a possible race where during didEnterForeground, signalCache is updated on the main queue while events can be pushed and popped from other queues.

• If the SDK gets one of the following response codes when transmitting events, it will not attempt to send these events again (they will be permanently dropped): 400, 401, 403, 404, 413, 422, 501, 505.