[JENKINS-76200] Start stopped instances in attemptReconnectIfOffline + extensive debug logging

This commit adds both a fix and extensive debug logging to diagnose and resolve
JENKINS-76200 where existing offline Jenkins nodes with stopped EC2 instances
are never restarted, causing jobs to wait indefinitely.

Root Cause Analysis:
When an existing Jenkins node is offline with a stopped EC2 instance, Jenkins
does NOT call start() on the retention strategy. Instead, it periodically calls
check() which calls attemptReconnectIfOffline(). The original implementation
only attempted reconnection for RUNNING instances, leaving stopped instances
in limbo forever.

Changes Made:

1. Added extensive debug logging throughout EC2RetentionStrategy:

   • start(): Logs when called, instance state during startup
   • check(): Logs when executing and when calling attemptReconnectIfOffline()
   • attemptReconnectIfOffline(): Logs state, offline status, connecting status,
     all decision branches, AWS API calls, and any errors

   All debug logs are tagged with [JENKINS-76200] for easy filtering.

2. Modified attemptReconnectIfOffline() to start stopped instances:

   • Check if instance is STOPPED or STOPPING
   • If offline, call AWS startInstances() API to start the instance
   • Log the start operation and result
   • Return immediately after starting (don't attempt connection yet)
   • On next check() cycle, instance should be PENDING or RUNNING
   • Existing reconnection logic handles RUNNING instances

3. Added necessary imports:

   • software.amazon.awssdk.services.ec2.Ec2Client
   • software.amazon.awssdk.services.ec2.model.StartInstancesRequest

Testing Instructions:

1. Deploy this build to Jenkins with an offline node that has a stopped instance

2. Trigger a job that requires the offline node

3. Check Jenkins logs for [JENKINS-76200] messages showing:

   • check() being called
   • attemptReconnectIfOffline() detecting STOPPED state
   • AWS startInstances() being called
   • Instance transitioning to PENDING/RUNNING
   • Successful connection

4. Verify in AWS console that the instance actually starts

Expected Behavior:

• Logs should clearly show the flow from check() -> attemptReconnectIfOffline()
• Instance should start in AWS when detected as STOPPED
• Node should eventually come online and job should execute