[Task Manager] Optimize status field output for health api

[chrisronline]

Relates to #101505

There are some points of confusion related to the task manager health API response that we should discuss and potentially fix.

1. The runtime health status is always OK, even though we set the overall status to Error based on a runtime metric

2. The workload health status is always OK, even though we set the overall status to Error based on a workload metric

3. In [Task Manager] Log at different levels based on the state #101751, we are logging a warning if the p99 runtime drift - maybe we should set the status to warning for the runtime bucket when this happens?

[elasticmachine]

Pinging @elastic/kibana-alerting-services (Team:Alerting Services)

[gmmorris]

The idea was that each section is updated independently and so when you get the overall health it would look at the constituent parts and reflect the overall health so that:

1. If any part is in Error, then the whole thing is in Error
2. If any part say's it's OK but it's not fresh enough (the last update was OK but for some reason hasn't updated in 5 minutes) then the overall API returns an Error satte
3. It would be clear which part was in error and which is ok

We can obviously throw that idea out 🤷 but that's where the idea comes from.