fix: intrinsic alarm crash in the storage example (the "Hmm?" exception)

[gralin]

Discovered while running AnotherStorageImplementation and clicking the device in YABE: as soon as an object's intrinsic limit alarm fired, the device crashed with:

System.Exception: Hmm?
  at System.IO.BACnet.Serialize.Services.EncodeEventNotifyData(...) Services.cs:1224
  ...
  at BaCSharp.NotificationClass...SendIntrinsectEvent...

Root causes

1. Core — EncodeEventNotifyData had a leftover placeholder throw new Exception("Hmm?") in the EVENT_CHANGE_OF_VALUE default branch. Replaced with a meaningful NotImplementedException (matching the sibling switch one level up).
2. Example — AnalogObjectEvent reported analog high/low-limit alarms as EVENT_CHANGE_OF_VALUE with no payload, which the encoder can't serialize. It now sends EVENT_OUT_OF_RANGE with the proper payload (exceeding value, status flags, deadband, exceeded limit) — the correct BACnet event for an analog limit alarm.
3. Example robustness — the notification is sent on a thread-pool thread; an unhandled exception there took down the whole device process. Wrapped it in try/catch so a failed send is logged instead.

Verified

• Core + example build; 119 tests pass.
• EVENT_OUT_OF_RANGE with payload now encodes cleanly; an unset CHANGE_OF_VALUE throws the new meaningful NotImplementedException instead of "Hmm?".

Result: clicking the device in YABE now shows a proper Out-Of-Range alarm instead of crashing.