[Stack Monitoring] ESS enables ingest_pipeline dataset for all deployment by default

[miltonhultgren]

If the customer also uses a lot of Ingest Pipelines this dataset can grow big really quickly, costing the customer on storage, regardless of if they're using this data or not.

This issue is partially hidden by elastic/beats#41850, causing this data to end up in metricbeat-* rather than together with the monitoring data in .monitoring-*.

The customer can drop this data using an ingest pipeline or adding a delete phase to the ILM policy for the Metricbeat indices.

The main blocker for changing this behaviour to be off by default is that we need to be able to configure this per deployment in ESS.