Cherry-pick of PR #24055 to 7.x branch. Original message:

What does this PR do?

This PR fixes 2 things.

First one is an issue introduced in #23886, this caused that VarSkipKey removed key from Action which was then stored using Ack and Store. This action was input-less so after restart agent was not running anything and it thought it;s ok because preserved state id is same as the one from fleet.
This one I discovered without filing an issue

Second one is bug described in #23720. The issue here is that agent creates a server with own counter, set to 1. But service already has a config with state counter. So if we change log level and restart an agent counter is 1 which == to the counter in a service and agent thinks everything is up to date.
Fix is that we take counter from a service and if it's > 0, which means services already retrieved some configuration we force reload of config by incrementing our local counter to what_service_has + 1

Why is it important?

Fixes bugs

Checklist

• My code follows the style guidelines of this project
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• I have made corresponding change to the default configuration files
• I have added tests that prove my fix is effective or that my feature works
• I have added an entry in CHANGELOG.next.asciidoc or CHANGELOG-developer.next.asciidoc.