Edit: after #2188 there are close to no documents covering ScyllaDB Manager deployment or task scheduling, only a brief overview.

What should the feature do?

We should refactor the pages of our docs related to Scylla Manager integration.

/kind documentation

What is the use case behind this feature?

Currently, the pages of our docs related to Scylla Manager are all over the place. There is an aggregated page Deploying Scylla Manager on a Kubernetes Cluster covering the deployment and task scheduling. Then, in an entirely different place, there's Restore from backup covering the manual procedure for backup restore.

I propose we refactor this part of our docs by creating a separate subtree dedicated to Scylla Manager integration, covering the points outlined in the Acceptance Criteria in separate pages.

Possibly also additional documents (or integrated into the above) for manager integration with multi-datacenter ScyllaDB clusters.

Alternatively, we can wait with this until we implement proper isolation and managed deployments.

Anything else we need to know?

The purpose of this issue is first and foremost to serve as a place for suggestions and discussion related to Scylla Manager integration docs.

Acceptance Criteria

• current architecture of Scylla Manager in k8s clusters
• deploying Scylla Manager in k8s clusters (both single-DC and multi-DC)
  • specify that in both manual and automated multi-DC, the manager instance may exist only in the control plane cluster (in case of automated) or one chosen cluster (in case of manual)
• scheduling backups and repairs, including the user providing credentials to object storage
• running one-off tasks, suspending and resuming tasks, what operations are allowed manually with sctool
• manual restore from backup procedure
• opting out of a ScyllaCluster global manager registration
• setting resource requests/limits for an agent according to the user's needs