Guide Request: Operational guide for hpa based on custom metrics

[LukoJy3D]

Is your feature request related to a problem? Please describe.
After setting up prometheus-adapter for Querier hpa to work with custom metrics, it wasn't easy to find proper scaling values to make that scaling work and be similar to KEDA. A few years back, @periklis shared excellent prometheus-adapter configurations in #5339, yet a lot of time passed, and it would be interesting to see what setup Grafana uses.

Loki has undergone many changes over the last few years, so it would be nice to refresh https://grafana.com/docs/loki/latest/operations/autoscaling_queriers as well. At first glance, it's not clear which approach (custom metrics or keda) offers less operational complexity.

Describe the solution you'd like
Up-to-date operational guides for HPA based on custom metrics and scaling with KEDA from someone operating at a larger scale.