Fix: Respect IReconnectRetryPolicy in disconnect loops

[NickCraver]

Fixes the case where we loop from Connecting -> BeginConnectAsync -> OnDisconnected and the next heartbeat triggers a reconnect again due to the backoff thresholds resetting. This ultimately causes the backoff to not be respected.

Here, we're doing 2 things:

• Upping the max exponential backoff window to 60 seconds which better handlers clients at scale and mass DDoS cases.
• Only resets the reconnect counter on the first disconnect, not doing so again until we've successfully reconnected and heartbeated again, so backoffs are properly respected.