Fix race condition in maintenance event rebind completion

Problem

Race condition during MOVING notifications when processed in the same Netty buffer read. The rebind completion check was performed too early, causing timeouts when MOVING notifications and command responses appeared in the same buffer.

Solution

• Moved rebind state check from beginning to end of decode cycle in CommandHandler
• Only mark rebind complete when command stack is actually empty after processing
• Added debug logging for rebind progress tracking

Changes

• CommandHandler.java: Moved rebindInProgress check to end of decode() method
• MaintenanceAwareConnectionWatchdog.java: Enhanced logging in address supplier

Impact

Fixes timeouts during endpoint rebind operations on single proxy policy.

Fixes: CAE-1386