JENKINS-72224 Prevent deadlock during frequent calls to onConfigurationChange()

Unexpected deadlock added on

• [JENKINS-72224] Trigger node updates when refreshing platform labels #1389

I was missing that the onConfigurationChange was also calling the refreshModel. It caused deadlock on busy instance with many nodes.

This PR restores the original code from refreshModel that just probe for label.

The saveNodeLabel it's now just called on the preOnline phase just after labels are collected.

The use case is really for ephemeral agent that connect frequently and to ensure those labels sees the new node.

I think there is a almost no chance that platform labels changes when a node is modified or when an other node connect. According to https://javadoc.jenkins.io/hudson/slaves/ComputerListener.html#onConfigurationChange() it's called every time a node is added or removed

Checklist

• I have read the CONTRIBUTING doc
• I have referenced the Jira issue related to my changes in one or more commit messages
• Unit tests pass locally with my changes
• No spotbugs warnings were introduced with my changes
• I have interactively tested my changes

Types of changes

What types of changes does your code introduce?

• Bug fix (non-breaking change which fixes an issue)