Ability to define the AutoScalingGroup GlobalTimeout

For login/compute nodes that takes longer than 10 minutes to complete, the instance is killed after initialization due to heartbeat-timeout for the aws autoscaling group lifecycle hooks.

We operate a cluster with SELinux enabled, so relabeling takes a very long time(on our efs and lustre), and would have to manually update the ec2 autoscaling group with a higher timeout so that our nodes stay alive after being spun up.

Start by tracing where the AWS Auto Scaling Group lifecycle hook and heartbeat timeout are configured. Verify how initialization time is passed into that configuration, then confirm that a user-defined GlobalTimeout lets SELinux relabeling and other long initialization complete without the instance being killed.

Written by the indexing model from the issue text.